SwissFEL Injector Conceptual Design Report · in July 2010 during the assembly phase. PSI – SwissFEL Injector Conceptual Design Report 3 ntroduction1 I 4 2 Mission of the test facility

PSI Bericht Nr. 10-05

Juli 2010

SwissFEL Injector Conceptual Design ReportAccelerator test facility for SwissFEL

Editor

Marco Pedrozzi

With contributions from:

V. Arsov

B. Beutner

M. Dehler

A. Falone

W. Fichte

A. Fuchs

R. Ganter

C. Hauri

S. Hunziker

R. Ischebeck

H. Jöhri

Y. Kim

M. Negrazus

P. Pearce

J.-Y. Raguin

S. Reiche

V. Schlott

T. Schietinger

T. Schilcher

L. Schulz

W. Tron

D. Vermeulen

E. Zimoch

J. Wickström

Cover figures descriptions

Overview in the tunnel of the SwissFEL injector test facility

in July 2010 during the assembly phase.

PSI – SwissFEL Injector Conceptual Design Report 3

1 Introduction 4

2 Mission of the test facility 5

3 Performance Goals 6

4 Accelerator layout 8

4.1 Overview 8

4.2 Electron source 9

4.3 RF systems 11

4.4 Diagnostics 19

4.5 Laser system 25

4.6 Girder system and alignment concept 29

5 Beam dynamics simulated performances 30

5.1 Initial conditions 30

5.2 Injector optimizations 32

6 Commissioning and operation program 36

6.1 Definition of the main commissioning phases 36

6.2 Operating regimes 37

6.3 Software applications and data storage 37

7 Controls 38

7.1 Requirements 38

7.2 Control System Structure and Architecture 39

7.3 Special Subsystems 40

8 Radiation Safety and Protection System 42

8.1 Introduction 42

8.2 The SHIELD11 computer code 42

8.3 Special Subsystems 42

8.4 Shielding calculations 43

8.5 Lateral shielding 43

8.6 Shielding of the silo roof 44

8.7 Shielding of the 0.3 GeV electron beam dump 44

9 References 48

Table of contents

4 PSI – SwissFEL Injector Conceptual Design Report

1 Introduction

The SwissFEL X-ray FEL project at PSI, involves the de-

velopment of an injector complex that enables operation

of a SASE FEL at 0.1–7 nm with permanent-magnet

undulator technology and minimum beam energy, i.e.,

a cost-effective way to obtain laser-like radiation in the

X-ray spectral region. In order to extensively study the

generation, transport and time compression of high

brightness beams and to support the component devel-

opments necessary for the XFEL project, PSI is pres-

ently planning a flexible 250 MeV injector test facility.

The injector pre-project is motivated by the challenging

electron beam requirements necessary to drive the

SwissFEL accelerator facility. First, the source and the

injector linac determine the minimum achievable emit-

tance of the electron beam after full acceleration and

therefore the maximum beam brightness at the undula-

tor. Second, the shot to shot stability of the FEL radiate

power, his arrival time and spectrum depends strongly

on the injector fluctuations which must be kept below a

reasonable level. More details about specifications for

the Free Electron Laser project can be found in the

SwissFEL conceptual report [1].


Generally the injector test facility must fulfill two main

purposes. Firstly it must provide a tool to benchmark the

performances predicted by the simulation codes and to

demonstrate the feasibility of a compact Free Electron

Laser. Secondly it will be the ideal platform to develop

and test the different components/systems and optimi-

zation procedures necessary to operate later SwissFEL.

In more detail we can list the missions of the facility as

follow:

• The facility provides the necessary background to

analyze the generation and propagation of high bright-

ness beams compatible with the specification of

SwissFEL. The diagnostic system is designed for a

careful comparison with the Beam Dynamic Simula-

tions:

– The machine is designed for optimal emittance

compensation at 250 MeV (invariant envelope match-

ing technique)

– A compression chicane is implemented at 250 MeV

and optimized for minimum emittance growth.

– Monitoring of the beam position and size with a

resolution ≤10 μm along the entire machine (inte-

grated BPM-screen monitors).

– The diagnostic line after the compression chicane

allows measurements of projected and sliced pa-

rameters (phase space characterization).

• During the injector commissioning and optimization

the operator crew will elaborate, optimize and automize

the commissioning and operation procedures which

will later be transferred to SwissFEL:

– QE measurement procedure

– GUN-Laser-Solenoid alignment procedure

– Beam Based Alignment procedure

– Projected parameter measurements

– Slice parameter measurements

– Emittance optimization procedure (invariant envelope

matching)

– Compression optimization (CSR studies)

– Orbit feed back, bunch length feed back, energy feed

back

• The short and bright beam bunches delivered by the

injector are used to support component developments:

the key hardware components are developed and

tested in order to reach the required functionality for

SwissFEL:

– Gun tests (gun developments, photo cathode surface

preparation)

– RF (phase and amplitude stability – femtosecond

synchronization)

– Diagnostics (BPM and screen resolution, slice

parameters, bunch length monitor, arrival time

monitor, charge monitors)

– Laser (longitudinal and transverse shaping, wave

length emittance tuning & stability)

– Mechanical stability and alignment concept (Girder

concept and support systems)

– Controls (expert panels and operations panels)

– Fast linac tuning and restore procedures

• Establish commissioning/operation experience and

consolidate the know how for SwissFEL commissioning

and operation.

2 Mission of the test facility


The scientific objectives of the injector facility can gener-

ally be summarized in terms of beam performance,

component functionality, machine stability and reliability.

The primary objective of the installation is to explore the

generation of high brightness beams according to the

requirements of SwissFEL. The beam parameter range

to be reached and explored with the injector is directly

based on SwissFEL baseline specifications.

The feasibility of a compact SwissFEL has been analyzed

via start to end beam dynamics simulations including

the FEL undulator lines. To establish a robust baseline

concept the injector has been designed starting from

the experience accumulated with advanced RF photo-

cathodes similar to the LCLS photo-cathode source [2].

Nevertheless the option of integrating later the innova-

tive Low Emittance Gun (LEG) presently under investiga-

tion at PSI [3] has been kept. The typical parameter

3 Performance Goals

SwissFEL Parameter Range

Operation mode Long pulse Short pulse Ultra short pulse

Single bunch charge (pC) 200 10 10

rms spotsize of gun drive laser (μm) 270 100 100

FWHM pulse length of gun drive laser (ps) 9.9 3.7 3.7

Peak current at the cathode (A) 22 3 3

Maximum number of bunches / pulse 2 2 2

Minimum bunch spacing (ns) / pulse 50 50 50

Normalized projected emittance (mm.mrad) 0.65 0.25 0.45

Normalized slice emittance (mm.mrad) 0.43 0.18 0.25

rms slice energy spread (keV) 350 250 1000

Peak current at the entrance of undulator (kA) 2.7 0.7 7

Beam energy for 0.1 nm (GeV) 5.8 5.8 5.8

rms photon pulse length at 0.1 nm (fs) 12 2.1 0.3

Number of photon at 0.1 nm (×1·109) 24 1 4

Saturation length (m) 45 50 30

Bandwidth (x1·10-4) 3.1 2.5 5

Maximum repetition rate (Hz) 100 100 100

Table 1: Parameter range in the SwissFEL baseline.

3 Performance Goals PSI – SwissFEL Injector Conceptual Design Report 7

range of the SwissFEL is summarized in Table 1. Par-

ticularly challenging is the extremely low transverse

emittance necessary to guarantee the electron beam

transverse coherence during the lasing process at

0.1 nm.

The results of the beam dynamic simulations for the

injector test facility are briefly summarized in Table 2.

According to our analysis the test facility should deliver

a beam well matched to the FEL specifications shown

above with reasonable margin to take into account pos-

sible emittance dilution effects along the main linac.

Due to the relatively low energy and the initial beam

manipulations the FEL injector up to the first compres-

sion chicane is one of the most critical section of the

overall accelerator complex. The RF phase fluctuations,

the alignment errors and mechanical vibrations of the

components and the synchronization systems can induce

large shot to shot fluctuations of the FEL radiated wave

length and power. Energy, arrival time, orbit and peak

current must thus be kept within severe tolerances to

comply with the experimental requirements at the FEL

photon beam lines. These tolerances imply not only

extremely stable RF and synchronization systems but

also the capability of diagnosing with sufficient resolution

and correcting dynamical errors. For this reasons the

Injector will be used as platform for the diagnostic de-

velopments and to test the feed back methods to be

later applied to SwissFEL. Table 3 illustrates briefly

typical tolerances for SwissFEL that will be explored

during the operation of the injector test facility and which

imply a staged R&D effort to reach the required final

functionalities of all components [4].

Typical tolerances for SwissFEL

Goal arrival time stability (fs) 20

Goal relative energy stability at 2.1 and 5.8 GeV

5·10-4

Goal peak current stability at the undulator (%)

5

Injector tolerances

Photo cathode laser pointing stability (μm) <10

Photo cathode laser energy stability (%) 0.84

Photo cathode laser jitter (fs) 55

Charge stability (pC) 1.68

S-band RF phase stability (deg) 0.022

S-band RF amplitude stability (%) 0.04

X-band RF phase stability (deg) 0.033

X-band RF amplitude stability (%) 0.2

Main Linac tolerances

C-band RF phase stability (deg) 0.315

C-band RF amplitude stability (%) 0.1

Undulator line quadrupole alignment (μm) 3

Orbit stability in the undulator section (μm) 1

Table 3: Typical tolerances for the SwissFEL.

Parameter Input parameter and simulated performances at 250 MeV

Charge (pC) 200 100 10

Laser diameter on cathode (μm) 1080 857 400

Laser pulse length FWHM – uniform distribution (ps) 9.9 7.9 3.7

Peak current on cathode (A) 22 14 3

Thermal emittance (mm.mrad) 0.195 0.155 0.072

Core normalized slice emittance before/after BC (mm.mrad) 0.32/0.33 0.213/0.213 0.078/0.078

Normalized projected emittance before BC (mm.mrad) 0.35 0.233 0.095

Normalized X/Y projected emittance after BC (mm.mrad) 0.379/0.35 0.268/0.233 0.104/0.096

RMS Bunch length before BC (ps) 2.8 2.23 1.05

RMS Bunch length after BC (fs) 193.3 117.6 33.2

Peak current after BC (A) 352 285 104

RMS beam size on OTR in 3rd FODO cell (μm) 55 35 14

Table 2: Input beam parameters at the electron source and simulated output at 250 MeV.


4.1 Overview

As shown in Figure 1 the electrons emerge on the left

side of the facility from an S-band RF photoinjector. A

booster linac based on normal conducting S-band RF

technology increases the energy up to approximately

270 MeV generating simultaneously the necessary en-

ergy chirp for the magnetic compression. In front of the

magnetic chicane a fourth harmonic X-band cavity,

phased for deceleration, linearizes the longitudinal phase

space for optimal bunch compression suppressing the

formation of peaks in the head and tail of the current

profile. The last drift section is dedicated to the beam

characterisation.

The design of the beam-optics of the 250 MeV linac

relies on the technique of so-called emittance compen-

sation [5, 6, 7], where space-charge forces are care-

fully used via a beam refocusing to control the natural

evolution of the emittance in the first few meters of the

accelerator.

It is important to appreciate that the dynamics of high

brightness beams in the initial phase of acceleration are

highly dominated by space-charge forces. With respect

to this point the SwissFEL injector doesn’t differ sub-

stantially from other injectors designs. A good measure

4 Accelerator layout

Figure 1: General layout of the 250 MeV injector facility.

to evaluate the dominance of space-charge is the so-

called laminarity parameter ρ [8], which reflects the ratio

in influence on the beam-dynamics between space-charge

and emittance

(4-1)

where I, σxy, γ, εn denote the current, the beam radius,

the Lorentz factor associated with the beam energy, and

the normalized beam-emittance, respectively. The lami-

narity parameter is scaled with the Alfèn current IA ≈

17 kA. The high brightness photoinjectors considered for

FEL or collider applications are usually largely space

charge dominated. The parameters quoted in Table 4 for

three different FEL photoinjectors lead to ρ factors at

low energy well above unity, indicating a predominant

influence of space charge on the beam dynamics. From

the theory of emittance compensation [9], it follows that

the space-charge forces can be controlled, provided

sufficient care is given to the design of the accelerating

system.

SwissFELXFEL [10]

LCLS [11]

Bunch charge (nC) 0.2 0.01 1 1

Peak current (A) 22 3 50 100

Gun peak gradient (MV/m) 100 100 60 120

rms beam radius at matching point (mm)

0.3 0.1 1 0.41

Beam energy at matching point (MeV)

6.6 6.6 6.6 6.4

Normalized slice emittance (mm.mrad)

0.33 0.08 0.9 0.9

Beam brightness (A/mm2mrad2)

404 940 123 247

Laminarity parameter 38.4 9.9 130.5 45.1

Table 4: Laminarity parameter of the central slices for different

photoinjectors estimated at the injection in the first accelerating

structure following the electron source.

It is also important to keep in mind that in these regimes

the required optics of the accelerator is dominated by

the space-charge density of the electron beam. This has

two important consequences:

• The beam-optics design of the linac must be optimized

for a so-called working-point, which is defined by the

bunch-charge and temporal profile of the bunch at each

location along the machine. For the 250 MeV linac,

the starting point is defined by the electron source

parameters as specified in Table 2. However, the linac

design presented in the following part of this chapter

may be adapted to other source parameters. Gener-

ally, such an adaptation requires both a readjustment

of the fields and location of solenoids, accelerating

structures, etc. For practical reasons the design is

chosen such that a large range of working points can

be adapted without changing the location of the hard-

ware.

• In the space charge dominated regime, it is important

to maintain the circular symmetry of the electron beam

since this provides an equal space-charge density in

the horizontal and vertical plane. By contrast, a non-

circular beam requires a machine design that provides

a separate solution for the horizontal and vertical plane

simultaneously. For this reason, all beam elements

maintain the circular geometry at low energy where

space-charge forces become dominant. Therefore the

focusing in the linac section includes only solenoids.

Quadrupoles are permitted in special combinations at

the high-energy end of the linac, where space-charge

forces are less dominant.

4.2 Electron source

To establish a robust baseline concept for the SwissFEL

we decided first to concentrate our investigations on an

electron source topology based on state of the art RF

photo-injectors benefiting from the progress accumu-

lated in the two last decades. Particularly encouraging

in this field are the recent results achieved at LCLS [12]

which complies with the specification of the SwissFEL.

For the commissioning of the 250 MeV injector, an exist-

ing RF photocathode electron source will be used until

a new electron source with higher performances, pres-

ently under development, will be ready (LEG [13] or

advanced RF-photo injector). According to simulations

(see Section 5) and recent experimental measurements,

an injector based on a standard RF photo-cathode can

already reach a slice emittance near to 0.3 mm.mrad.

The operation of the 250 MeV test injector will start with

the CTF3 gun number 5 [14, 15], kindly lent by CERN to

PSI. This RF-photo injector is a 2.5 cell standing wave

cavity that was originally developed for high current

operation at the CLIC test facility. The first half cell is

overmoded in order to store enough RF energy to com-

pensate beam loading effects, while the rear wall is

4 Accelerator layout PSI – SwissFEL Injector Conceptual Design Report 9


shaped in order to provide some additional focusing

(Figure 2). Two RF coupling apertures are symmetrically

located on the external wall of the last cell. The sym-

metric coupling will suppress the dipole component of

the accelerating field in the last cell but no racetrack

geometry was adopted to minimize the residual quadru-

pole component. A RF field probe pick up and frequency

tuners are available on each cell.

Figure 2: Cross section of the CTF gun number 5.

Following delivery, the cavity has been characterized at

PSI to verify the RF parameters and the field profile. The

measured values are summarized in Figure 3.

The high power behavior of the cavity has been tested

during one week in the SLS linac bunker (Figure 4). The

nominal gradient of 100 MV/m was reached with a

forward input power of 21 MW, and a temperature of the

cavity of 47 °C [16].

Figure 4: The CTF gun number 5 in the SLS linac bunker during the

high gradient tests.

According to the start to end simulations (see paragraph

5) the CTF 3 gun performs sufficiently well to deliver

beam charges and emittances within the SwissFEL

specifications. Nevertheless this source is not fully

compatible with the operation modes foreseen for the

FEL facility. First of all, the tuners overheat for repetition

rates above 10 Hz and the overall cooling is not compat-

ible with 100 Hz operation; secondly the effect of the

quadrupole kick in the last cell on the final emittance

has not been evaluated; and finally the cathode plug

system in the first half cell is a weak point concerning

dark current. For these reasons a procurement program

for a new photo-injector has been initiated. The goal is

to produce a new gun to be delivered by mid of 2011.

The design strategy is based on an existing state of the

art photo injector (LCLS and PHIN [17]). The cooling

scheme will be modified to allow operation at a repetition

rate of 100 Hz, and the mechanical assembly optimized

in order to facilitate the cathode replacement for easier

cathode surface treatments investigations.


Figure 3: RF parameter of the

CTF3 gun number 5 measured

at PSI. The field profile has

been measured with the bead-

pull technique.


4.3 RF systems

4.3.1 General description

The general layout of the RF system is shown in Figure

5. The accelerator starts with the S-band RF-photoinjec-

tor described in section 4.2. A booster linac composed

of four S-band structures, 4 m long, provide the required

acceleration and energy chirp for the magnetic compres-

sion. A 12 GHz cavity is placed in front of the compres-

sion chicane for linearization of the longitudinal phase

space. Two additional S-band deflecting cavities are

implemented for diagnostics purposes in order to char-

acterize the longitudinal phase space and slice param-

eters of the beam.

For maximum flexibility each accelerating structure will

be powered by a dedicated power station. This choice

allows a fine control of the RF focusing (gradient optimi-

zation) for optimal invariant envelope matching avoiding

any cross talk between cavities. SLED compressors have

not been considered for the initial phase to suppress

any additional complication. The modulator and klystron

can nevertheless be operated in a short pulse mode for

a maximum gradient test (25 MV/m) and long pulse

mode (45 MW) for a possible later installation of a SLED

compressor system.

4.3.2 Accelerating structures

Booster linac

The booster linac consists of four constant gradient 2π/3

travelling wave accelerating structures separated by a

short drift where corrector, BPM and screen monitor are

situated. The cavity RF design is based on the Linac II

DESY structures and was re-optimized, compromising

between maximum shunt impedance to decrease the

power requirement at 25 MV/m and short range wake-

field considerations. For small correction of the trans-

verse focusing each cavity is surrounded by four 70 cm

long solenoid magnets.

The design parameters of the S-band accelerating struc-

tures are summarized in Table 5.

Operating frequency at 40 °C (MHz) 2997.912

Phase advance 2π/3

Maximum total length flange to flange (mm) 4300

Number of cells 122

Operating temperature (°C) 38 – 42 ±0.1

Input coupler: Double-feed type racetrack geometry

Output coupler Double-feed type racetrack geometry

Nominal accelerating gradient (MV/m) 20

Maximum accelerating gradient (MV/m) 25

Shunt impedance per unit length (MΩ/m) 62

Peak power for 25 MV/m (MW) 57

Filling time (μs) 0.955

RF pulse length at 60 MW (μs) 1.2

Maximum RF pulse length at 45 MW (μs)

4.5

Maximum pulse repetition rate (Hz) 100

Material of the cavity Cu-OFE

Material of the flanges, supports and cool-ing connectors

316LN

Concentricity tolerance (μm) ± 150

Design pressure of the cooling channels (Bar)

16

Table 5: RF parameters for the booster cavities of the 250 MeV in-

jector. The nominal accelerating gradient depends on the position

of the structure and ranges from 10 to 21 MV/m as described in

paragraph 5.2.


Figure 5: General layout of the Rf systems for the 250 MeV injector test facility. The operational modes of the klystrons are described in

Table 8.


The iris design has been optimized introducing an el-

liptical shape to reduce the maximum surface electric

field (Figure 6). The iris radius “a” ranges from 12.695 mm

at the cavity entrance down to 9.31 mm at the cavity

end. The input and output coupler cavities have been

symmetrized with dual feed and racetrack geometry. The

ratio between maximum surface electric field and ac-

celerating field is described in Figure 7.

0 20 40 60 80 100 1200

10

20

30

40

50

60

Cell number

Eac

c, Esu

rf (

MV

/m)

Eacc

Esurf

Figure 7: Accelerating field and maximum surface field along the

booster cavity for 25 MV/m accelerating gradient.

The X-band harmonic cavity

The X-band harmonic cavity is a modified SLAC H75

structure. This structure is being developed in collabora-

tion with CERN where the cavity will be tested for high

gradient applications within the CLIC high gradient R&D

program. Two cavities will be manufactured for PSI and

two for the FERMI project. The cavity is a constant-gra-

dient un-damped travelling wave structure with 5π/6

cell-to-cell phase advance and 72 cells, (including the

matching cells) [18]. The active length is 750 mm.

The cavity will be operated typically at peak gradient of

21.2 MV/m resulting in a total decelerating voltage of

approximately 16 MV. The iris aperture (see Figure 8)

ranges from 4.993 mm down to 4.107 mm. With such

a small aperture this cavity will dominate the impedance

budget of the entire linac. Misalignments larger than

10 μm can generate large transverse kicks with severe

consequences on the final emittance. For this reason

cell numbers 36 and 63 have been radially coupled to

four wave guide systems (Figure 9) to extract dipole

mode signals which will give an indication about the

cavity misalignment and optimum beam positioning.

Figure 6: Detail of the iris design of the booster cavities and 3D model of the wave guide system at the input coupler (only the first cell is

shown here).

Figure 8: Gradient, iris radius and cell radius along the harmonic X-band cavity.

Parameter Dimension

p (mm) 33.333

t (mm) 5

r (mm) 10

a (mm) 12.695 9.31



A mode launcher will be used to couple the RF power

into the cavity and a similar structure will be imple-

mented at the end of the structure to couple out the

residual power (Figure 10).

The manufacture process will require a machining preci-

sion below 2 μm. According to the experience accumu-

lated at CERN and SLAC in this field, such specification

can be reached with modern temperature and vibration

stabilized precision-machining tools

Figure 10: X-band cavity 3D model of the mode launcher and EM simulation of the input/output matching cells with electric field amplitude.

Figure 9: Special cell design with output ports for dipole mode

monitoring.

No. of Cells 1

Frequency (MHz) 2997.912

Nominal/Maximum deflecting voltage (kV) 300/440

Nominal/Maximum input power (kW) 90/200

Repetition rate (Hz) 100

Filling time ( μs) ~0.8

Nearest mode – different polarity (MHz) 50

Length flange to flange (m) 0.1

Quality factor – Q0 ~15000

Shunt Impedance deflecting mode R (MΩ) 0.5

Aperture Beam pipe (diameter) (mm) 38

Working temperature 40°

Temperature range (°C) 35 – 50

No. Pickups 1

Table 6: Specification fort the low energy one cell deflector.

RF deflectors

The first deflecting cavity at low

energy will be a single cell pill

box cavity developed in collabo-

ration with INFN. A wave guide

port will be used as input cou-

pler and a small antenna as a

pick up. The RF parameters are

summarized in Table 6.



The high energy deflector is scaled from the structure

developed for the SPARC project [19]. It will be manu-

factured by INFN within a collaboration framework with

FERMI at ELETTRA and PSI. The cavity is a 5 cell stand-

ing wave structure (see Figure 11); the operating mode

is the TM110.π like hybrid mode. The RF parameters of

this structure are summarized in Table 7.

Figure 11: 3D model of the deflecting cavity designed by INFN for

the FERMI project and the PSI version with the tuners displaced

on the upper side.

No. of Cells 5

Frequency (MHz) 2997.912

Nominal/Maximum deflecting voltage (MV) 4.5/4.9

Nominal/Maximum input power (MW) 4.2/5

Operating mode πRepetition rate (Hz) 100

Filling time (μs) ~0.8

Nearest mode – different polarity (MHz) 50

Length flange to flange (m) 0.5

Quality factor – Q0 ~15000

Shunt Impedance deflecting mode R (MΩ) 2.4

Aperture Beam pipe – diameter (mm) 38

Working temperature 40°

Temperature range (°C) 35 – 50

No. Pickups 2

Table 7: Specification for the 5 cell RF deflector to be installed at

250 MeV.

4.3.3 Power systems

The power modulators provide the high voltage pulsed

power for the klystron amplifiers. These different sourc-

es are characterized by their peak power, pulse width,

and high voltage and current capabilities. The three dif-

ferent klystron configurations in the FEL injector and

accelerators, together with the maximum electrical pa-

rameters of the modulator power sources are given

below in table 8.

Modulators that provide this high power are connected

to the klystron loads through low leakage inductance

and low stray capacitance high voltage step-up pulse

transformers. The flat top of the voltage pulses produced

are approximately equal to the RF pulse width of the

klystron. The modulator energy efficiency is mainly de-

termined by the rise and fall times of the voltage pulses,

which will be approximately 1 μs and 2 μs (10 to 90 %)

respectively.

The primary energy storage in each modulator can be

either a pulse forming network (PFN) or switched ca-

pacitors. The PFN line type modulator method (Figure 12

and Figure 13) provides a fixed pulse width voltage pulse

having a source impedance that is matched to the klys-

tron impedance. The stored energy in the PFN is made

just sufficient to satisfy the klystron load requirements.

In this case the high voltage switch on the primary side

of the pulse transformer needs to be only an ON device

such as a Thyratron or a Sylicon Controlled Rectifier

(SCR) assembly.

Figure 12: Line type modulator block diagram.

100 100

441±0.3

ø36±0.05



In the second method (see Figure 14) where a capacitor

discharge source is used, the stored energy has to be

much larger than that required by the klystron load to

ensure a voltage flat top region with a small percentage

voltage drop. Additional protection is needed in this case

to protect the klystron from receiving excessive energy

from the storage capacitor in the case of a gun arc. This

type of modulator commonly uses an IGBT solid-state

ON-OFF switch to define the voltage pulse duration. Since

these devices operate in a lower voltage region (3 to

6 kV) the step-up ratio of the pulse transformer needs

to be a higher ratio than when a Thyratron switch is used

(30 to 40 kV), and consequently requires a special trans-

former technology to ensure that the rise and fall times

mentioned above can be achieved.

Figure 13: Typical scheme of a PFN modulator.

Figure 14: Electrical scheme of a solid state modulator.

3 GHzRF gun

3 GHz Linac – mode 1 to 3 12 GH3 GHz

DeflectorUnit

RF Peak Power 30 30 45 60 50 7.5 MW

RF Average Power 15 4.5 22.5 9 5 3.75 kW

Peak power klystron 71.4 69.7 104 146 107 15.7 MW

Average power klystron 53.1 19.6 65.2 42.5 26.5 10.2 kW

Klystron voltage range 266 261 307 351 410 148 kV

Klystron current range 268 267 340 416 262 106 A

Pulse flat top length 5 1.5 5 1.5 1 5 μs

Max pulse length (at 76%) 7.4 3.9 7.4 3.9 3.4 7.4 μs

Repetition rate 10–100 10–100 10-100 10-100 Hz

Flattop flatness (dV) <±0.2 <±0.2 <±0.2 <±0.2 %

Rise rate (at 50 % voltage) 170/270 250–360 300-420 148 kV/μs

Fall rate (at 50 % voltage) 170/270 250–360 300–420 70–148 kV/μs

Amplitude stability <±0.02 <±0.2 <±0.02 <±0.02 %

Pulse to pulse time jitter <5 <5 <5 <5 ns

Pulse length Jitter <±10 <±10 <±10 <±10 ns

Table 8: Modulator operational parameters of the klystron amplifiers.



Both these types of modulator have been investigated

from the point of view of performance specification,

reliability, maintainability and cost.

PFN modulators using conventional Thyratron switch

technology are already in operation at the LEG test facil-

ity and the SLS machine, with known reliability and run-

ning experience. The main weakness of this approach,

especially at high repetition rates, is the relatively short

lifetime of the Thyratron switch (15’000 to 20’000

hours), which substantially contributes to the operation

costs and reliability. The pulse behavior is fixed by the

number of cells in the PFN and can be modified only by

changing the capacitor configuration inside the modula-

tor rack. The size of the components and the mechanical

assembly is adapted to the relatively high voltages of

the PFN line, fixing a physical limit to the compactness

of the HV cabinet.

A low power solid–state modulator has been purchased

for the LEG test facility introducing a new technology at

PSI. This very compact device has been tested showing

an extremely fast installation and commissioning proce-

dure. The modulator is composed of a series of low

voltage (~1 kV) switching modules connected in parallel

to the step up transformer (see Figure 14). In general

the solid state modulator tested at PSI presents the

following advantages:

1. Compactness

2. Fast installation (only three main connection for

electrical power, cooling and controls)

3. Modularity (the same switch modules can be stacked

in different configurations, less spare components

needed)

4. Reliability (70’000 hours without failures recorded

by the suppliers, a few modules can fail without

perturbing the modulator operation)

5. Adjustable pulse length

6. KHz repetition rate

At a comparable investment cost and based on the

positive experience obtained so far, PSI has decided to

use the solid state option for the injector test facility.

The 3 GHz and 12 GHz modulators requires respectively

voltages of 300 and 450 KV for a modulator peak power

exceeding 140 MW. The design needs therefore a large

amount of parallel operated solid-state IGBT devices with

a high ratio step-up pulse transformer to handle the

power that a PFN modulator with a single Thyratron or

SCR assembly would handle. The new modulator concept

consist in a “single rack” amplifier units integrating the

tank, the water distribution and the power supplies for

the klystron solenoids and heater (see Figure 15).

Figure 15: The high power modulator K2 from the company SCAN-

DINOVA (10-60MW; 200-450 kV).

The klystron loads for the modulators are high gain RF

power amplifiers that usually have a narrow bandwidth

(10 MHz) at the fundamental frequency. The Thales

TH2100 klystron shown below (Figure 16) is a typical

example.

The RF output pulses from each klystron are transmitted

via an RF window that is conditioned to operate with an

SF6 gas pressure of about 2 bars. This electro-negative

gas enhances the voltage hold-off capability in the pres-

ence of high forward and reflected power occurring in

phase together at the window. If this should occur the

used gas is recovered in a closed circuit system and the

window volume is flushed with clean gas and re-filled.

Figure 16: High-power

klystron amplifier TH2100.

As mentioned above the shot to shot RF phase and

amplitude stability are particularly critical because they

can potentially generate large FEL fluctuations (arrival

time, power, wave length). Assuming a stable distribution



network, the RF phase and amplitude are mainly driven

by the modulator voltage stability. To meet the FEL

specifications (~0.01 deg RF) the relative voltage stabil-

ity must be of the order of 10-5. This will require an effort

in ameliorating the capacitor charging scheme which

usually reaches values one order of magnitude larger

than the specified ones. To overcome this problem we

plan at the injector facility, and in collaboration with the

modulator suppliers, an R&D program to improve the

performances of the presently commercial available

hardware. Recent encouraging measurements of the

achievable phase stability have been carried out at ELET-

TRA in the framework of the FERMI FEL project [20]. The

experimental data reproduced below in Figure 17 indi-

cates that a conventional PFN based S-band power plant

is able to reach shot to shot stabilities below 0.1° peak

to peak. A slow feed back system is nevertheless

needed to correct slow phase fluctuations which might

be induced by thermal drifts of the power plant compo-

nents.

Figure 17: Phase stability of a conventional S-band power plant at

ELETTRA (Courtesy of G. D’Auria).

4.3.4 Low level RF systems and synchronization concept

The purpose of the Low Level RF (LLRF) control system

is to synchronize and regulate the RF fields in the ac-

celerating cavities throughout the machine. For the injec-

tor, three different types of accelerating structures have

to be controlled, namely

• 2.5 cell RF-photo injector, fundamental frequency,

3 GHz (standing wave structure)

• The booster accelerating structures, 3 GHz (traveling

wave structure)

• The harmonic accelerating structure, 12 GHz (traveling

wave structure).

Errors in amplitude and phase produce energy modula-

tions, particularly if the beam is accelerated off-crest,

which result in timing jitters from pulse to pulse after

the compression chicane. The requirements in terms of

amplitude and phase stability are given in Table 9. The

table summarize the required RF tolerances needed at

different stages of the injector commissioning and for

the SwissFEL operation. The figures are based on pre-

liminary tolerance studies for SwissFEL [3] and on further

optimizations reported in the SwissFEL CDR [1].

Gun alone

Injector without BC1

Injector with BC1

Swiss-FEL

S-band amplitude stability (%)

0.1 0.05 0.04 0.04

S-band phase stability (deg)

0.3 0.05 0.022 0.022

X-band amplitude stability (%)

0.2 0.2

X-band phase stability (deg)

0.2 0.033

Table 9: Estimation of the amplitude and phase rms stability re-

quirements (rms) for the RF accelerating structures at different

phases of the injector commissioning and for SwissFEL.

* timing stability of the driving laser for photoemission of 0.1 ps

to 0.2 ps.

To fulfill these requirements several stages are foreseen.

The basis for stable RF input powers to any of the ac-

celerating structures are stable klystron modulators (see

section 4.3.3). The promising results which are shown

there indicate the need for a slow, pulse to pulse regula-

tion of the input amplitude and phase. In case of sys-

tematic repetitive errors within the few microseconds

long RF pulses a fast feed forward system could in ad-

dition reduce these perturbations. These tasks can best

be achieved by digital feedback and feed forward sys-

tems. The latter especially has the potential to use

adaptive algorithms in order to compensate systematic

pulse to pulse distortions. It is therefore foreseen to

implement the LLRF system as a digital system which

is well integrated in the accelerator control system. A

generic overview of the RF cavity control is depicted in

Figure 18.

The Low Level RF Electronics concept is based on down

converting any RF signals (3, 12 GHz) to a common in-

termediate fif frequency of 46.84 MHz which is the 64th

sub-harmonic of 3 GHz. These signals are digitized by

16 bit ADCs sampling with 125 MHz (24th sub-harmonic



of 3 GHz) synchronously with the Master Oscillator. On

state-of-the-art FPGA platforms (Xilinx Virtex5) the in-

phase and quadrature-phase components of the RF

vectors are extracted by non-IQ sampling [21]. Digital

pulse to pulse feedbacks and adaptive feed forward

systems to compensate repetitive systematic distortions

will maintain the very tight amplitude and phase stabil-

ity requirements of the individual RF cavities. Vector

modulators will control the drive signal to the high

power klystrons. The I/Q input signals of the vector

modulators are generated by 16 bit DACs driven by the

same 125 MHz clock as the ADCs.

A required phase stability of 0.01 deg at 3 GHz corre-

sponds to a temporal RF stability of approximately 10 fs

making the overall synchronization scheme particularly

challenging. The pulse to pulse RF stability will mainly

be determined by the high voltage pulse to pulse stabil-

ity of the modulators which has to be on the order of

10-5. Although PFN based modulators have already

reached this stability [22] the applied solid state based

modulators have still to demonstrate this performance.

Concerning the synchronization, all local oscillator, ADC

and DAC clock signals are derived from the 3 and 12 GHz

reference signals which are generated locally at each

RF station by PLL systems locked to the distributed

reference signal of 214 MHz. A temperature stabilized

LLRF distribution system and an optical synchronization

system will be installed in parallel (Figure 19). While the

more robust RF-based system will guarantee machine

operation with sub-50 fs peak to peak stability, the more

sophisticated optical timing distribution will be tested

during the injector operation aiming for sub-10 fs timing

jitter. Details of the distribution system can be found in

[23].

Figure 19: Simplified layout of electrical and optical reference dis-

tribution and temperature controlled cable pipe for the electrical

distribution.

Figure 18: Generic LLRF system layout for the accelerating structures. For a standing wave cavity, with no field probe is available, the cavity

field vector has to be calculated based on the measurements of the forward and reflected power.



All drive signals for the klystrons and local oscillator

frequencies for IQ detection are either generated by

standard frequency multiplication/division techniques

or by conversion of the optical signal of the synchroniza-

tion laser link to an electrical RF signal. The Optical

Master Oscillator (OMO) produces laser pulse trains with

repetition frequency fR = 214.4 MHz which generate a

spectrum of all harmonics n·fR at the output of a high

bandwidth photo diode. Selecting the appropriate har-

monics by bandpass filtering provides the desired RF

frequency as a drive or local oscillator frequency. High

frequencies above 4.5 GHz are most likely generated

from the laser pulse spectrum by means of dielectric

resonator oscillators (DRO) [24].

The RF gun cavity integrates field probes to extract the

signal by means of bandpass filters. If a field probe

should be omitted in order not to disturb the radial sym-

metry of the RF cavity the accelerating fields in the RF

gun have to be calculated from the forward and re-

flected powers requiring precise calibration of these

signals [25].

The frequency tuning of the RF cavities during operation

will be accomplished by a temperature tuning system.

The digital LLRF system will determine the detuning of

the cavity from pulse to pulse. The LLRF systems have

to analyze this information, calculate the necessary

tuning change and set the new set-points for the tem-

perature accordingly. A digital interface between the

LLRF and the tuning system will provide the necessary

communication. Standard cooling control units typically

regulate the temperature within 0.1°C which corresponds

to ~5 KHz for an S-band cavity. This causes a change in

RF phase of about 2.6° (@3 GHz). A pulse to pulse LLRF

feedback system is therefore compulsory to compensate

temperature induced phase drifts and adjust the RF

cavity phase to the required level of stability. Stabilization

of a 1.6-cell RF gun operating at 2.86 GHz to less than

0.1° in phase and 0.1% in amplitude has been achieved

so far with a similar digital LLRF approach [26, 27].

Following the modular approach for the LLRF system

presented above only the analog front ends and the

vector modulator have to be adapted to the different

accelerating frequencies. The digital signal processing

unit including the software will be mainly generic for all

LLRF systems.

4.4 Diagnostics

4.4.1 Introduction – General Strategy for Beam Diagnostics

The 250 MeV X-ray Free Electron Laser (X-FEL) Injector

Test Facility at the Paul Scherrer Institute will focus on

the production, transport and diagnosis of high bright-

ness electron beams. Using this facility the acceleration

concept will be carefully investigated in order to demon-

strate the feasibility of an economic X-FEL facility at PSI.

Electron beam diagnostics will be designed to support

the demonstration of the challenging SwissFEL Injector

beam parameters and will thus need to provide high

performance with excellent reliability and reproducibility.

In addition, prototype developments for SwissFEL diag-

nostic systems will be made at the 250 MeV Injector

Test Facility in order to re-engineer the chosen solutions

for robustness and cost-effectiveness and to improve

their performance and reliability at the same time. In-

formation about absolute beam and lattice properties

will be required for most of the measurements, while in

some cases the online monitoring of relative beam pa-

rameters will support stable operation of the accelerator

sub-sections. All diagnostic devices need to provide

single-shot information. Measurement of projected and

“sliced” beam parameters will be provided in the gun

and pre-injector region (up to 8 MeV) and after bunch

compression at full beam energy of 250 MeV. With regard

to these general requirements and boundary conditions,

the overall design concept for the electron beam diag-

nostics will aim for redundant delivery of information,

accurate calibration of monitors and cross-check as well

as cross-calibration of diagnostic devices (whereever

possible). A modular configuration of components will

provide sufficient robustness and reliability and permit

at the same time enough flexibility to incorporate improve-

ments, which may be necessary to validate the final

accelerator performance and the SwissFEL beam param-

eters. Most of the (standard) monitors will be based on

designs which have already been successfully utilized

in the SLS (Swiss Light Source) pre-injector linac and at

other linac-based FEL facilities. Adaptations to the spe-

cific needs and requirements of the SwissFEL project

(especially the low charge and short bunch mode) as

well as potential improvements regarding performance,

simplicity of operation and reliability of the monitors will

be incorporated in the designs at an early stage.



The personnel and competences required to provide

state-of-the-art diagnostics for the 250 MeV Injector and

the SwissFEL project will of course profit from the exist-

ing expertise of the PSI Diagnostics Section, specifi-

cally in the fields of beam position measurements and

beam-based feedback systems. Experience with opera-

tional systems (such as e.g. the SLS digital beam posi-

tion monitors [28], the fast orbit and multi-bunch feed-

backs [29, 30] and the (optical) emittance monitor at

SLS [31]) as well as new developments (such as the

cavity BPMs and the “intra bunch train feedback” for the

European XFEL) will support the efficient realization of

high-quality monitors in these areas. Some of the com-

petences for key diagnostics in linac-based light sources,

however, have to be built up with the start of the 250MeV

Test Facility. Particularly longitudinal (bunch length and

arrival time) and sliced diagnostics as well as timing and

synchronization on a sub-100 fs (later sub-10 fs) level

need to be efficiently introduced in order to realize high

resolution monitors and operational diagnostics systems

in due time. While basic concepts in these areas have

been developed at PSI in the course of Doctoral studies

and first results have been achieved at the SLS pre-in-

jector LINAC [32] and for the SLS Femto-slicing project

[33], a lot more effort and manpower has to be invested

to reach the state-of-the-art at other LINAC-based radia-

tion sources (such as FLASH and LCLS) and to provide

the required measurements in a reliable manner and as

a valuable support of accelerator commissioning and

development.

Layout of Test Injector Diagnostics Sections

Following the general layout of the 250 MeV Injector Test

Facility, the diagnostics sections can be divided in three

sectors:

• the RF photoinjector,

• the S-band LINAC and bunch compressor and

• the high energy (250 MeV) diagnostics branch to ana-

lyze all final beam properties of the injector.

4.4.2 Low Energy Electron Beam Diagnostics

The low energy (<8 MeV) beam transport line between

the electron gun and the first S-band accelerator struc-

ture has a length of almost 3 m, providing sufficient

space for diagnostic monitors. Moreover, it is foreseen

to test different electron gun settings over a wide range

of beam parameters at the 250 MeV Injector Test Facil-

ity – starting with the CLIC 2 ½ cell RF gun, followed by

Figure 20: RF-gun with the first 3 m drift section and the beginning of the first accelerating structure.



an optimized PSI RF gun design. Thus, the possibility of

a complete characterization of the projected and “sliced”

electron beam phase space has been integrated in the

low energy diagnostics line.

A set of four optical screen monitors allows the control,

monitoring and optimization of the transverse beam

envelope matching into the S-band injector linac. Due to

the low beam energy YAG: Ce crystals will be used for

the profile measurements. The readout system of the

screen monitors consist of two selectable optical branch-

es: the first, with a large useable area on the order of

the beam pipe diameter, for electron beam optimization

and halo detection providing medium spatial resolution

of ~25 μm/pixel (projected pixel size). The second, with

a small useable area of interest of up to 5 mm (maxi-

mum), providing high spatial resolution ~5 μm/pixel

(projected pixel size) for beam measurements. A detailed

description of the prototype screen monitor and optics

design is given in [34]. A transverse (projected) emittance

scan is possible by using the three quadupoles of the

low energy line in combination with the 4th screen

monitor directly in front of the 1st S-band linac structure.

A set of horizontal and vertical slit masks as well as a

“pepperpot” mask has been integrated in the combined

resonant stripline BPM/screen monitor station directly

in front the injector linac providing an alternative method

to the quadrupole or gun solenoid scan for the determi-

nation of transverse beam emittance in case of a space

charge dominated beam (200pC operation mode). Imag-

ing of the corresponding beamlets is possible by using

the drift length of ~4.5m to the next screen monitor

station, located between the 1st and the 2nd S-band linac

structure.

Two beam position monitors (BPM) – a button type (due

to the restricted available space) and a 500 MHz reso-

nant stripline [35] – are integrated in the low energy

beam line for electron beam trajectory measurements.

While the button type BPM pick-up provides only limited

resolution of ~100μm (especially in the low charge

operation mode) at the exit of the gun solenoid, the

resonant stripline BPM will be able to determine the

transverse beam position with ~10 μm single-shot reso-

lution [36]. By using the low energy corrector magnets,

a beam position feedback for the entrance point of the

1st S-band linac structure will be implemented.

A coaxial Faraday cup, a wall current monitor and an

integrating current transformer (ICT) [37] will be used to

monitor the bunch charge and the charge profile within

the gun RF pulse. While the coaxial Faraday cup and the

wall current monitor allow the observation and optimiza-

tion of possible dark current from the RF gun, the ICT

can be used for monitoring the quantum efficiency of the

photo cathode and for the implementation of a possible

bunch charge feedback by keeping the gun laser inten-

sity constant.

An electro-optical bunch length monitor [38] using the

method of spectral decoding will be installed in a com-

bined screen monitor/EO monitor station (CF40 diag-

nostics cube) at 2249 mm downstream of the photo

cathode. Using a GaP bi-refringent crystal in the UHV

beam pipe and an Yb-doped fiber laser at 1030 nm

wavelength [39], the Coulomb field of the electron bunch

can be probed with sufficient time resolution (<1 ps) to

monitor the rectangular electron bunch profile with its

700 fs rise and fall time and 9.9ps (FWHM) duration. In

addition, this monitor allows the comparison of the

longitudinal electron beam profile with the auto-correla-

tion measurement of the longitudinal laser pulse profile

to optimize the laser beam transport line from the laser

hutch to the photo cathode in the accelerator bunker.

The performance and stability as well as reproducibility

criteria of this EO-monitor will be tested in the low en-

ergy gun region in order to provide effective single-shot

bunch length information with similar systems at higher

energies (250 MeV) and at shorter (sub-ps) electron

bunch lengths before and after the bunch compressor.

Likewise, an electro-optical electron bunch arrival time

monitor (BAM) will be implemented in the low energy

beam transport line between the RF gun and the injector

linac to monitor the beam arrival time with respect to a

stabilized optical synchronization link using a commercial

Er-doped fiber laser with extremely low (3.3 fs) jitter [40].

A high bandwidth (>15 GHz) pick-up consisting of four

button electrodes (2 mm diameter) and four ridged

waveguides has been designed to generate a fast

(<100 ps slew rate) voltage pulse of about ±10 Vpp in

both operation modes (button pick-up for 200 pC and

ridged waveguides for 10 pC) of the Injector Test facility.

While the electro-optical BAM in the gun region will pro-

vide information about the photo-cathode laser timing

stability, the performance of two more BAMs, respec-

tively in front of and behind the bunch compressor, will

monitor the energy and RF amplitude (and phase) stabil-

ity of the injector accelerator structures. In addition to

the beam arrival time measurement, it is planned to

compare the distribution of the 214 MHz reference sig-

nal (laser pulse repetition rate) over stabilized optical



links with a more conventional distribution using a RF

microwave master oscillator and RF microwave compo-

nents, in order to decide on the synchronization scheme

for the final reference distribution of the SwissFEL.

The momentum and momentum spread of the electron

beam can be measured through dispersion created by

a dipole magnet (30° bending) in a spectrometer arm.

Absolute momentum measurement is given by the ge-

ometry and calibration of the dipole and the subsequent

drift length (600 mm). In order to maximize the momen-

tum resolution of the spectrometer, the dispersive

contribution to beam size should be large compared to

the emittance contribution. This is achieved by use of

the quadrupole triplet in front of the dipole, which pro-

vides a sharp horizontal focus at YAG:Ce doped screen,

SME, and limits the beam size vertically. The setting for

the focus is not known a priori, since space charge ef-

fects will perturb the beam transport matrix transforma-

tion, and the focus will be hard to find on SME where

the beam is widened by dispersion. Therefore another

screen, SCR20, is located in the non-dispersive, straight

beam path with the same path length. The dipole MB-

ND10 is a gradient free rectangular magnet, which be-

haves like a drift horizontally, so a focus produced at

SCR20 and well observable there, will be moved to SME

after switching on MBND10.

A single cell S-band transverse RF deflector has been

designed in collaboration with INFN Frascati. More de-

tailed information about the design and the RF specifica-

tions are given in the RF section of this report. The de-

flector will be installed for measurement of the electron

bunch length as well as “sliced” (horizontal) emittance

and sliced energy spread at 1000 mm downstream of

the gun. The maximum deflecting voltage of 300 kV and

the drift space (1700 mm) to the screen SCR20 in front

of the 1st S-band accelerating structure will provide

~200 fs time resolution, which is sufficient to resolve

the 700 fs rise and fall time of the longitudinal bunch

profile in both operation modes (200 pC and 10 pC) of

the Injector Test Facility. A quadrupole triplet can be used

to increase the resolution of the longitudinal phase space

measurements.

4.4.3 Linac and Bunch Compressor Diagnostics

Between each of the S-band accelerator structures

combined BPM/screen monitor stations, consisting of

a 500 MHz resonant stripline BPM and a YAG:Ce and

OTR (optical transition radiation) screen, allows monitor-

ing the electron beam position and to visualize the

transverse electron beam profile. Two additional BPM/

screen monitor stations will allow matching and position

control of the electron beam through the X-band structure

located donstream of the injector linac. The screen

monitor station directly in front of the bunch compressor

can be used in combination with the upstream quadru-

poles to determine transverse beam emittance and Twiss

parameters for correct matching into the bunch compres-

sor. A beam position and angular feedback, which is

driven by the BPM readings in front of the bunch com-

pressor is used to provide stable beam trajectory condi-

tions for any given bunch compressor setting. Transmis-

sion through the S-band linac can be measured with an

ICT. An EO bunch length monitor applying the method of

spectral decoding allows the online control and monitor-

ing of the longitudinal charge distribution in front of the

bunch compressor. A set of two EO beam arrival time

monitors (BAM) – one in front of the bunch compressor

and one directly behind the bunch compressor – can be

used to determine the beam arrival time. The BAM sig-

nals can be used in a (slow) feedback loop to prevent

beam energy drifts and to stabilize the RF amplitudes

of the injector linac structures. The expected energy

resolution from the BAM measurements is ~4x10-5,

corresponding to <10 fs beam arrival time measurement

resolution.

Two of the standard 500 MHz resonant stripline BPMs

at the entrance and exit of the bunch compressor will

be used to monitor the energy and energy stability of

the electron beam. Although the BPMs are not located

at maximum dispersion in the center of the bunch com-

pressor (R56 of 430 mm) the expected position resolution

of 10 μm and the R56 of 90 mm at the location of the

BPMs will provide an energy resolution of ~1x10-4. The

BPM performance will be compared to the BAMs, in

order to determine the most suited energy and RF am-

plitude feedback configuration for later use in SwissFEL.

A pair of horizontal slits in the center of the bunch com-

pressor will be used to cut energy tails of the beam. A

screen monitor directly behind the slits provides informa-

tion about the energy spread induced by the off-crest

acceleration in the S-band linac and the linearity of the



energy chirp achieved by the X-band structure. A syn-

chrotron radiation monitor behind the third BC bending

magnet provides the same information in a non-destruc-

tive manner for online monitoring and possible feedback

applications. Coherent synchrotron radiation (CSR) in

the mm-wave and THz spectral range will be coupled out

from the fourth (last) bending magnet of the bunch

compressor. The integral intensity will be measured with

a Golay Cell or pyro-electric detector to control, monitor

and optimize the compression of the electron bunches.

The CSR-based compression monitor will also allow to

find the correct phase setting of the S-band LINAC struc-

tures and might be used to implement a feedback on

the RF phase settings of the linac.

4.4.4 High Energy Electron Beam Diagnostics

All electron beam parameters (projected as well as sliced

beam properties), which are relevant for validating the

RF Photoinjector performance and the acceleration (injec-

tor) concept for SwissFEL have to be measured at the

end of the Injector Test Facility at 250 MeV beam en-

ergy.

A 5-cell transverse RF deflecting cavity (TDS), which has

been redesigned by INFN Frascati to European S-band

frequencies for use at the FERMI and the SwissFEL

projects, will be installed in the diagnostics section

behind the bunch compressor. More detailed RF speci-

fications of this TDS are listed in the RF section of this

report. Since the main purpose of the SwissFEL Injector

Test Facility is focused on the optimization of components

and accelerator sub-systems to obtain a high brightness,

low emittance electron beam, a FODO lattice has been

chosen to allow the measurement of the projected (TDS

off) and “sliced (TDS on) horizontal emittance without

changing the beam optics settings. A matching optics

consisting of 5 quadupoles generates a vertical β-function

of 20 m in the TDS providing efficient vertical deflection

of the beam. Five more quadrupoles are used for match-

ing into the three FODO cells with a total horizontal phase

advance ψx of 165° (55° per FODO cell). The horizontal

β functions of 5m at the location of the 7 screen moni-

tor stations, which are distributed along the FODO cells,

provide a time resolution of ~20 fs respectively ~10 fs

assuming up to 4.5 MV deflecting voltage and the desired

beam emittances of 0.4 mm.mrad for 200 pC and 0.125

mm.mrad for 10 pC respectively. The TDS time resolution

corresponds to ~10 slices along the 185 fs (rms) long

bunch in case of the 200 pC operation mode, while in

the short bunch (10 pC) mode only 3 slices can be

visualized over the extremely short bunch length of

~30 fs (rms). A dipole magnet at the end of the diagnos-

tics line can be used as a spectrometer to measure the

projected and “sliced” energy distribution of the beam.

The left side of figure 21 provides a schematic view of

the 250 MeV diagnostics section including the horizon-

tal (βx) and vertical (βy) β-functions as well as the beam

dispersion ηx. The phase advances (ψx, ψy) along the

FODO lattice are shown on the right side of Figure 1.

Figure 21: β-functions and dispersion (left side) as well as phase

advances (right side) along the 250 MeV SwissFEL Test Injector

diagnostics line.

A total of 8 BPMs will ensure sufficient beam trajectory

control along the diagnostics section and a third ICT will

be used for beam charge measurement and determina-

tion of overall transmission through the 250 MeV injec-

tor. As already indicated in the previous paragraph, an

EO bunch length monitor applying the technique of spec-

tral decoding (≤150 fs time resolution at 200 pC bunch

charge) will be installed behind the bunch compressor

for online monitoring of the longitudinal bunch charge



distribution. Although, this device provides redundant

information compared to the TDS for the case of the

250 MeV test injector, it is foreseen to simplify the de-

sign and to improve its robustness for online operation

in the background in order to provide a non-destructive,

single-bunch longitudinal profile measurements behind

the 1st bunch compressor of SwissFEL for its 200 pC

operation mode.

4.4.5 Summary of the proposed Electron Beam Diagnostics

A summary of the diagnostics components is given in

Table 10 through 12. It follows the previously introduced

partition of diagnostics sections in a low energy branch,

the linac segments and a high energy branch.

Device Name # of Devices Measured Property Resolution

Button beam position monitor 1 Beam position σavg/5 (100 μm)

Res. stripline beam position monitor 1 Beam position 10 μm @10–200 pC

Optical screen monitor (OSM) 4 Beam profile/positionEnergy /energy spread

~10 μm~10-4

“Pepperpot” / slits 1 Emittance <100 nm rad

Integrating current transformer (ICT 1 Charge, transmission <5 pC (<1 pC)

Faraday cup (FC) 1 Time structure/charge <200 ps / 10%

Wall current monitor (WCM) 1 Time structure/charge <200 ps / 10%

Bunch length monitor (BCM) 1 Bunch length using Electro Optical Sampling

>500 fs (over 20 ps)

Bunch arrival time monitor (BAM) 1 Arrival time using EO-detection, based on optical timing and synchronization

<10 fs

Deflecting cavity 1 Longitudinal current profile and slice parameters

<200 fs


Beam position monitor (BPM) 3 Beam position 10 μm @10–200 pC

Optical screen monitor (OSM) 3 Beam profile/position ~10 μm


Stripline beam position monitor (BPM) 15 Beam position 10 μm @10–200 pC

Optical screen monitor (OSM) 15+5 Beam profile/position ~10 μm

Synchrotron radiation monitor (SRM) 1 Beam profile/positionEnergy spread

~10 μm~10-4

Integrating current transformer (ICT 2 Charge, transmission <5 pC (<1 pC)

Bunch arrival time monitor (BAM) 1 Bunch arrival time <10 fs

EO bunch length monitor (BLM) 1 Bunch length ~150 fs

Bunching monitor (BM) 1 Bunching using coherent FIR radiation relative measurement

Transverse RF deflecting cavity 1 Sliced beam properties and bunch length

~20 fs (200 pC)~10 fs (10 pC)

Table 10 low energy (8 MeV) diagnostics devices.

Table 11: diagnostics devices in between the linac accelerator structures.

Table 12: high energy (250 MeV) BC and diagnostics section devices.



4.5 Laser systemA copper photo-cathode is presently the base line elec-

tron source candidate fulfilling the SwissFEL injector

requirements in terms of beam brightness. For the com-

missioning of the 250 MeV injector a laser system ca-

pable of generating 200 pC of photo-electrons is re-

quired. A minimum energy per pulse of 100 μJ at 266 nm

is required to produce 200 pC of photoelectrons from a

copper surface assuming a QE of 10-5. Such pulse ener-

gies can be achieved with presently available Ti:Sa

amplifier systems. In addition, Ti:Sa laser systems pro-

vide pulses as short as 100 fs so that it becomes pos-

sible to generate a uniform longitudinal profile (square

profile) with a rise time of few 100 fs by frequency domain

pulse shaping. The design, installation and operation of

such laser system are not trivial. A collaboration with

the Laser Group of the University of Bern has been initi-

ated for UV pulse characterization techniques [41].

The emitted bunch charge distribution is almost a direct

image of the longitudinal and transverse profile of the

laser pulse. The influence of the laser shape on emit-

tance preservation is well known. Some ideal profiles

known as “waterbag” profiles would ultimately linearize

the intra-bunch space charge forces [42] so that emit-

tance compensation scheme would be applicable. De-

velopment of novel laser pulse shaping techniques will

be done in collaboration with a French company (Fastlite).

A Ti:sapphire based amplifier system with subsequent

frequency conversion stages is the most promising start-

ing point for R&D activities to reach the required gun

laser performance (Table 13). Ti:sapphire systems are

commercially available and have reached a high techno-

logical maturity due to their widespread use in research

laboratories throughout the world. In addition, many di-

agnostics tools and pulse shaping techniques are avail-

able for the near-IR/visible spectral region and frequen-

cy-tripling of the fundamental laser wavelength allows

easy access to the UV.

Conventional Ti:sapphire amplifiers, however, suffer from

spectral narrowing due to the limited gain bandwidth

yielding spectra of 30–40 nm (FWHM) at the mJ level

making wavelength tuning impossible. The scheme pre-

sented here has the potential to overcome this limitation

and offers in addition enhanced pulse energy stability

and direct UV pulse shaping. This should help to produce

electron bunches with lower emittance at the gun.

Laser specifications

pulse energy 200 μJ

central wavelength 250–300 nm

bandwidth 2-3 nm

pulse repetition rate 100 Hz

laser spot size on cathode (s) 0.1–0.27 mm

pulse rise time <0.7ps

pulse duration (FWHM) 3–10 ps

longitudinal pulse form various

transverse pulse form Uniform

laser to RF phase jitter <100 fs

UV pulse energy fluctuation <0.5 % rms

pointing stability on cathode <1% ptp

Table 13: Gun laser characteristics for SwissFEL.

4.5.1 General Layout

The laser system (Figure 22) consists of 4 subsequent

amplifier stages. As seed laser, a Rainbow oscillator

(Femtolaser, Inc.) delivering 380 mW is used, pumped

by a 5 W Verdi (Coherent, Inc.). To improve the long-term

performance of the oscillator a beam pointing stabilizer

has been installed on the pump beam consisting of a

4-quadrant photodiode and a piezo-driven mirror mount.

The pre-amplified (10μJ) and temporally stretched

(≈200 ps) pulse seeds the regenerative amplifier. An

acousto-optic programmable gain control filter (Mazzler,

Fastlite Inc. [43]) is situated in the regenerative cavity

and is used as an adaptive spectral filter to counteract

spectral narrowing during amplification. Broad spectra

of up to 100 nm (FWHM) are subsequently amplified in

two sequential multi-pass amplifiers and finally com-

pressed to 20 fs, 18 mJ. Six identical Q-switched, fre-

quency-doubled Nd:YAG pump lasers (Centurion, Quantel,

Inc.) have been chosen to pump the various amplifier

stages with a total pump power of 120 mJ.

The compressed pulses are frequency-converted from

the near-IR to the UV by second harmonic generation

(SHG) in a β-barium borate (BBO) crystal (type I, 0.5 mm,

ϕ=29.2°, φ=90°) and subsequent sum-frequency gen-

eration (SFG) in a second BBO crystal (type I, 0.5 mm,

ϕ=42°, φ=90°). The fundamental temporal pulse shape

is optimized for highest UV conversion efficiency by the

grating based compressor and a Dazzler located after

the pre-amplifier. At this point the Dazzler is used to



compensate for third-order dispersion to avoid detrimen-

tal satellite pulses.

Temporal pulse forming will be done by a low-loss prism-

based pulse stretcher followed by an acousto-optical

programmable dispersive filter (UV-Dazzler, Fastlite Inc.)

while transverse homogenization will be realized by se-

lecting the core part of the intensity profile by a spatial

mask [44] which is relay-imaged onto the cathode sur-

face.

4.5.2 Pump laser cluster

Each of the Centurion pump lasers provides 20 mJ

@532 nm at a repetition rate of 100 Hz. The large

number of six identical pump lasers is uncommon for

the performance of the presented amplifier system. In

principle one single powerful pump laser could provide

the required pump energy. The present design, however,

has been a consequence of stability constraints of exist-

ing commercial amplifier systems. To fulfill the stringent

stability requirements of the gun laser we have chosen

the following approach. First, compact diode-pumped

Q-switched lasers (Centurion) from Quantel [45] have

been chosen since they provide lowest rms energy fluc-

tuations (<0.4% rms) at 532 nm (Figure 23).

Second, to increase stability further the number of pump

lasers have been increased to six since pump-laser in-

duced energy fluctuations can be significantly reduced

by mixing different pump sources. In principle, the net

noise of N pump lasers will be decreased by a factor of

�

N compared to a pump laser providing the same

amount of pump energy. The employment of six pump

lasers reduces the noise floor by a factor of ≈2.5. Pre-

liminary measurements show, that the rms stability of

the amplified pulse (0.34% rms) is in fact better than

the performance of one individual pump laser (0.4% rms)

(Table 14).

Figure 22: Schematic drawing of Ti:sapphire gun laser system.

Figure 23: Pulse energy fluctuation of an individual Centurion pump laser over 120 sec (rms ≈0.4%).



4.5.3 Wavelength selection

Control of the central laser wavelength (i.e. the laser

photon energy) is helpful to produce low-emittance elec-

trons [46]. Electrons leaving the metal surface by photo-

emission have a kinetic energy Ekin = hν − φeff, and thus

a thermal emittance εth ∝ Ekin 1/2 arising from the mismatch

of the photon energy hν and the cathode work function

φeff. The total emittance may be reduced by adapting the

laser photon energy to the net work function of the

cathode material.

Despite the large gain curve of Ti:sapphire crystals

(600–1100 nm) conventional amplifier systems provide

spectra of only 30–40 nm centred around 800 nm due

to gain narrowing effects. In our system an acousto-

optical gain shaper (Mazzler) implemented in the regen-

erative amplifier helps to overcome such bandwidth-

limiting effects [47].

The Mazzler introduces a wavelength dependent loss

inversely to the Ti:sapphire gain curve in order to achieve

a flattening of the net gain over a large spectral range.

The resulting flat-top like spectrum reaches typically

≥80 nm bandwidth at FWHM (Figure 24). For pulse shap-

ing purpose, however, spectra of 15–25 nm (FWHM) at

the fundamental wavelength provide enough bandwidth

for the generation of flat-top UV pulses with a rise time

of <0.5 ps.

Wavelength selection is performed by the Dazzler. The

current amplifier scheme allows continuous variation of

the central wavelength within a range of 760 to 840 nm

with a spectral width of 25 nm (Figure 25). Pulse energy

and pulse stability (rms/peak-to-peak) of individual

spectral slices are independent of the central wavelength

as depicted in Table 14.

λ central (nm) 785 805 835 805

bandwidth (nm) 25 25 25 90

stability (%, rms) 0.39 0.36 0.39 0.35

stability (%, P2P) 2.2 2 1.7 2.8

duration (min) 2 2 2 2

pulse energy (mJ) 18.2 18 17.9 18.2

Table 14: Ti:sapphire stability for individual spectral slices.

Figure 24: Broadband amplification in Ti:sapphire thanks to acousto-optical gain control.

Figure 25: Wavelength-selected narrow-band amplification.



4.5.4 UV pulse shaping

The spatial and temporal shape of the amplified laser

pulse is nominally Gaussian. Electron beam dynamics

simulations indicate, that a flat-top like pulse shape

helps to generate uniform temporal and spatial electron

distribution resulting in a lower transverse emittance at

the gun. Since other simulations show that a Gaussian-

like electron distribution is less sensitive to timing jitter

in the undulator it is more likely that an intermediate

pulse shape will be optimal for best FEL performance

[48]. Therefore powerful pulse shaping tools are re-

quired.

The formation of arbitrary UV pulses, however, has been

non trivial. In the past indirect techniques have been

applied such as frequency-domain pulse shaping at the

fundamental wavelength (near IR) due to the lack of ap-

propriate shaping tools in the UV. The required conver-

sion process to the UV, however, can lead to strong

distortion of the original pulse shape due to the non-

linearity of the process.

Our approach for achieving high-quality arbitrarily shaped

UV pulses is based on direct UV pulse stretching and

shaping. This became possible very recently thanks to

the availability of an efficient acousto-optical pulse

shaper working in the UV (UV-Dazzler, Fastlite Inc.). This

device allows efficient phase-and amplitude control in

the wavelength range of 250–400 nm. With a resolution

of 0.1 nm and a deflection efficiency of up to 40 % at

250 nm the UV Dazzler should allow the formation of

powerful arbitrarily shaped UV pulses. To avoid intensi-

ty-induced nonlinear absorption in the KDP crystal the

UV Dazzler is seeded with UV pulses stretched to sev-

eral ps. For temporal pulse stretching we designed a

dedicated prism assembly shown in Figure 26.

Figure 26: UV pulse stretcher based on a double-prism sequence.

Efficient pulse stretching in the UV is challenging since

gratings, which are commonly used for pulse stretchers

in the near-IR, suffer from high losses in this wavelength

range. In fact, the expected energy throughput of a

grating-based stretcher would be <20 %. A prism-based

design seems more promising. Various glass substrates

from different suppliers have been tested at PSI in order

to optimize the energy throughput by minimizing inten-

sity-induced two-photon absorption but still offering a

large angular dispersion to keep the geometrical dimen-

sions of the stretcher small.

A prototype double-prism stretcher is currently being built

at PSI based on Corning 7980 HPFS glass substrates.

Ray-tracing calculations indicate that a distance of 3 m

is required to stretch the UV pulse from ≈100 fs to the

required 9.9 ps. The expected energy throughput should

be close to 80 %. The pulse length can be continuously

varied within 2–9.9 ps by adapting the geometrical dis-

tance between the two prism pairs.

We expect the UV-Dazzler / UV-stretcher combination to

be a powerful tool for arbitrarily temporal pulse shaping

providing the requested UV pulse quality. Temporal pulse

characterization will be performed by cross-correlation

with the fundamental laser pulse.

4.5.5 Laser-Room for the SwissFEL Injector

The amplitude stability of laser systems is highly depend-

ent on environmental conditions (temperature, air turbu-

lence, humidity). In order to provide the best environ-

mental quality for the laser system, the SwissFEL

injector will include a temperature and humidity control-

led clean room. The room provides a Class 10000 en-

vironment with a temperature stability of ±0.1°C. The

humidity is limited to a maximum of 50 % in order to

avoid condensation on warm optical components.

The analog and optical Master Oscillator also require a

stable environment. For this reason both will be in the

laser room.



4.6 Girder system and alignment concept

For the girder design we decided to take profit of the

experience accumulated with the SLS girder system. The

component positioning and alignment concept is based

on the SLS design [49, 50] while the girder material will

be mineral cast. For the alignment two rails machined

and polished with a precision of 100 μm over 4.7 m will

define the reference surface for a direct positioning of

all components (Figure 27). Fine adjustment of the

components on a given girder will be performed using

calibrated thickness blocks (shimming), the relative

position girder to girder will be adjusted using the girder

jacks.

Figure 27: Support system for the S-band cavity solenoids. In

green a section of the girder, in red the polished reference surfac-

es.

The basic design criteria are summarized in Table 15.

Eigenfrequency >50 Hz

High internal damping

Each component can be adjusted individually with minimum increments <50 μm

Universal generic alignment concept for all components

Repeatability of adjustments <0.01 mm for critical components

Easy to adjust the level of the girder after settling of the ground floor (without changing the horizontal position)

Cost efficient for SwissFEL

Pre-assembly of components on girders

Mechanical accuracy of components in relation to the girder, without adjustments, better than 0.1 mm

Table 15: Design criteria for the Girder system.

To achieve the required stability and to optimize the cost

the option of using mineral cast will be investigated in

the 250 MeV injector facility. This material has proper-

ties very close to granite and is relatively widely used to

manufacture stable support tables for precision machin-

ing tools. Until now it has never been systematically

studied for girder systems supporting accelerator com-

ponents.

The main advantages of mineral cast can be summarized

has:

• high damping properties expected, non homogenous

material

• high eigenfrequency as one solid block

• cost effective in mass production (no high temperature

processes)

– one wooden mould gives ~10 girders

– one steel mould gives +100 girders

• surfaces can be ground to precision of 0.1mm over

the length

• no local stresses introduces by machining

To avoid strong coupling with the ground motion particu-

lar care has to be taken to push the first resonant modes

above 50 Hz where the energy transfer from the ground

to the structure become less efficient and the absolute

ground displacements smaller. Table 16 shows typical

eigen-frequencies of such a girder.

Modus Frequency [Hz]

1 116.89

2 125.16

3 147.5

4 179.6

5 212.16

6 231.62

Table 16: Simulated resonant frequency for a mineral cast girder.

The picture illustrate the first vibration mode of the mineral cast

girder.



5.1 Initial conditions

For a realistic simulation and optimization of the electron

source it is necessary to evaluate the uncorrelated

intrinsic emittance at the emission from the cathode

surface. For photo emission this value depends on the

difference between the laser photon energy and the

effective potential barrier in the crystal bulk in the pres-

ence of an external electric field (Figure 28).

The assumptions on the initial intrinsic emittance are

based on the measurements performed recently with

photo cathodes at PSI [51] and SLAC [12]. The sum-

mary of these measurements are shown in Figure 29.

For the simulations presented here below we considered

a mean value between the intrinsic emittance measured

at PSI and with the LCLS gun. Since the intrinsic emit-

tance is the ultimate factor limiting the brightness of an

electron source it is important to have a deep under-

standing of the emission process and to clarify the

discrepancies between theory and measurements. A

recent analysis of the photo-emission process from

metal cathodes can be found in [52].

5 Beam dynamics simulated performances

Figure 28: Three step emission model and expression for the approximated intrinsic emittance.

(5-1)


Figure 29: intrinsic emittance measurements performed with the

LCLS RF photo injector (courtesy of D. Dowell) and at the PSI LEG

test facility with a diamond turned copper cathode. The dotted line

represents the theoretical value calculated by D. Dowell assum-

ing a perfectly flat cathode surface. The blue dot is the value used

for the simulation reported here below.

Note that the theoretical evaluation of the intrinsic emit-

tance discussed above doesn’t take into account the

quality of the cathode surface; In particular the surface

roughness can deteriorate sensibly the thermal emit-

tance and could explain the discrepancy between the

measured values and the theoretical ones. In the CTF3

gun the cathode can be relatively easy exchanged allow-

ing studies with respect to the cathode preparation

procedures.

Parameter 200 pC 10 pC

Kav (eV) 0.4 0.4

Intrinsic Emittance (mm.mrad) 0.195 0.072

Laser rms Size (μm) 270 100

Laser Rectangular Pulse FWHM (ps) 9.9 3.7

Laser Rise and fall Time (ps) 0.7 0.7

Peak Current (A) 22 3

RF peak gradient (MV/m) 100 100

Table 17: Beam parameters at the photo cathode for high and low

charge operation.

The initial beam parameters are summarized in Table

17. As discussed above, a smaller intrinsic emittance

can be obtained by decreasing the laser spot size at the

cathode. However for a given charge the dilution effects

due to longitudinal and transverse space charge restrict

our degree of freedom. Therefore the peak current and

gradient in the gun must be globally considered while

optimizing the laser spot size and pulse length. The laser

rise and fall time correspond to the new Titanium-Sap-

phire laser system described in the paragraph 4.5.

Figure 30: Summary of the optimized parameters for 200 pC operation and 100 MV/m at the cathode.



5.2 Injector optimizations

5.2.1 Invariant envelope matching

The injector start to end simulations have been described

in [53]. Since the FEL interaction is a local process re-

stricted to a co-operation length, only the electron bunch

slice parameters determine the FEL radiation properties.

Nevertheless for optimal performance of the facility it

is important to guarantee an optimal Twiss parameter

matching along the electron bunch and consequently a

good compensation of the projected emittance from the

beginning of the accelerator. To achieve this result the

beam is manipulated at low energy benefiting from the

emittance fluctuations induced by the space charge ac-

cording to the technique known as “invariant envelope

matching” [54, 55]. For this purpose a solenoid magnet

is placed just at the exit of the RF photoinjector to refo-

cus the beam in front of the first travelling wave accel-

erating structure. The exact positioning of the first ac-

celerating structure must be optimized and depends

mainly on the charge density and beam energy (for the

geometry see also Figure 20).

In the configuration studied here the invariant envelope

matching is practically completed after the first two ac-

celerating structures where the beam parameters are

frozen by the high energy. Figure 30 illustrates the typi-

cal parameters of the gun and the first two travelling

wave cavities for optimum operation with 200 pC and

100 MV/m at CTF3 gun cathode.

Figure 31: Normalized emittance and beam envelope

along the first 12 m of the injector.

Figure 32: Current profile and slice emittance (left) and mismatch parameter (right) at 130 MeV after two travelling wave accelerating

structures.



Figure 33: Illustration of the envelope matching sensitivity, (a)

±1% gradient error, (b) ±1 deg Rf phase error, (c) ±1% Solenoid

field error.

Figure 31 illustrates the evolution of the projected nor-

malized emittance and beam envelope up to 130 MeV

for the situation illustrated in Figure 30. The first ac-

celerating cavity is positioned 2.95 m from the cathode

at the optimum matching point.

The quality of the matching can be measured looking at

the slice emittance along the bunch and using the beam

mismatch parameter given by:

(5-2)

Where β0, α0, γ0 are the matched projected Twiss pa-

rameters and β, α, γ the slice Twiss parameters. If the

mismatch parameter is equal to 1 all slices have the

same Twiss parameters and the matching is perfect. In

general after emittance compensation in the injector the

mismatch parameter for the beam core should be kept

below 1.1. Figure 32 shows the mismatch parameter

and slice emittance at 130 MeV for the optimized con-

figuration described here above.

After fixing the relative position between cathode and

first accelerating structure for the nominal operation the

optimization procedure can be applied for different

charges. Table 18 shows a summary of the simulation

results for charges between 200 and 10 pC at 130 MeV.

Charge(pC)

Laser pulseFWHM(ps)

Ipeak, cathode

(A)

Laser σx,y

(μm

εintrinsic

(μm.rad)

εslice/εprojected

(μm.rad)

200 9.9 22 270 0.195 0.320/0.350

150 9.0 18 245 0.177 0.272/0.283

100 7.9 14 214 0.155 0.220/0.233

50 6.2 8.7 170 0.123 0.160/0.174

20 4.6 4.7 125 0.091 0.108/0.122

10 3.7 3 100 0.072 0.080/0.096

Table 18: Optimized parameters for different charges between

200 and 10 pC.

Note that the invariant envelope matching is relatively

insensitive to RF phase variations in the photo-injector,

while a significant modification of the matching conditions

can be observed when changing the gun gradient and

the solenoid focusing field of 1%. Figure 33 shows the

expected performance deteriorations if one modifies the

above mentioned parameters around the optimum set-

tings for the operation with 100 pC.



5.2.2 Overall Injector optimization

A schematic layout of the overall injector including a

possible upgrade with an undulator line is shown in

Figure 34. After performing the invariant envelope match-

ing the electrons are accelerated off crest in the S-band

traveling wave cavities up to ~275 MeV. This process

induces the necessary energy correlation along the

bunch for the longitudinal compression in the magnetic

bunch compression chicane (BC). In front of the BC an

12 GHz X-band cavity operated in deceleration mode is

used to remove the longitudinal non linearity introduced

by the RF during the previous acceleration stages.

The lattice and optic in the bunch compressor region

has been optimized to minimize the emittance growth

Figure 34: Full injector facility - optimization for 200 pC operation.

Figure 35: Compression chicane layout and optics from the compression chicane to the FODO cells.



due to CSR effects in the dipoles, while the optic in the

diagnostic section has been optimized to allow quick

switches from projected and slice parameters measure-

ments without adjustments of the transport line. Figure

35 shows the concept of the compression chicane and

the nominal optics. The transversal position of the cen-

tral dipoles will be adjustable to have maximum flexibil-

ity for the compression studies.

Since the large dispersion in the center of the BC in-

duces a transverse beam size of σx~5.5 mm the dipole

magnets must be designed with a large “good field”

region to avoid non linear dispersion and non linear fo-

cusing effects that could dilute the emittance after

compression [56]. The design retained for the injector

dipoles has been optimized using the 3D solver TOSCA

[57]. As shown on Figure 36 the expected relative de-

viation in integrated field at different transverse positions

defines a “good field” region with ΔBintegrated/B <5x10-5

extending over ±40 mm.

Figure 36: Compression chicane dipole, design for large “good

field” region.

Table 19 summarizes the expected beam characteristics

after compression. According to these simulations the

injector should deliver a beam fulfilling the SwissFEL

requirements. For the final SwissFEL injector we are

currently studying an increase of the injector energy up

to a maximum of 450 MeV to alleviate space charge

effects during compression.

Parameter 200 pC 10 pC

Energy (MeV) 255

Rms Bunch Length (fs) 193 33.2

Rms Projected Emittance (mm.mrad) 0.38 0.1

Rms Slice Emittance (mm.mrad) 0.33 0.078

Peak Current (A) 352 104

Table 19: Beam parameters after compression for high and low

charge operation.



The machine commissioning will consist of three main

phases. For all these phases the repetition rate in op-

eration with beam is limited to 10 Hz. In special test

modes the RF plants could be operated at 100 Hz with

the exception of the RF-gun (CTF3 gun #5) which is lim-

ited to maximum 10 Hz and the deflecting cavities lim-

ited to 50 Hz.

6.1 Definition of the main commissioning phases

Phase 1 is dedicated to the commissioning of the RF

photoinjector and the first 3 m transport line, down-

stream some additional screens are added for beam

diagnosis . The component assembly and the infrastruc-

ture preparation must allow early operation of the gun

while assembling the rest of the machine. For this pur-

pose the linac tunnel will be divided in two sections by

a movable concrete wall shielding the area downstream

of the gun region. Inside the gun bunker it is foreseen

to start, parallel to the gun commissioning, the early

conditioning of the travelling wave structures. Commis-

sioning of phase 1 will start with a Nd:YAG laser while

awaiting for the commissioning and optimization of the

more sophisticated Ti:Sa laser system. The Nd:YAG laser

will allow full commissioning of all components and

procedures, with the exception of invariant envelope

matching, which relies on the longitudinal charge distri-

bution provided by the Ti:Sapph laser.

For phase 2 the machine will be fully assembled with

the exception of the X-band harmonic cavity and the

bunch compressor. The beam will be accelerated down

to the diagnostic straight first without compression and

invariant envelope matching optimized for smallest emit-

tance. The machine and the measurements procedures

will be optimized avoiding the additional complications

due to the small aperture and high impedance of the

harmonic cavity.

The X-band cavity is required for phase 3 to perform

systematic studies of the bunch compressor. In order

to reach the high peak current at the undulator a two

stage compression is foreseen for SwissFEL. The first,

stronger, compression stage happens at the end of the

injector linac via a magnetic compression chicane. The

coherent and incoherent synchrotron radiation emitted

by the electron along the chicane dipoles and collective

effects like micro-bunching instability can seriously dilute

the beam emittance if not kept under control. The veri-

fication of the beam parameters as reported in Table 19

are the main goal of the beam dynamics scientific pro-

gram at this phase of the commissioning.

Main steps phase 1:

• RF conditioning CTF gun up to 100 MV/m

• Laser alignment on cathode (laser pointing stability,

laser-RF synchronization)

• Beam-based alignment of gun solenoid

• Schottky-scan to find laser reference phase, cathode

wquantum efficiency

• Dark current characterization

• Beam-based alignment of magnets

6 Commissioning and operation program


• Commissioning of the spectrometer: momentum and

momentum spread (RF phase scan)

• Commissioning BPMs: beam orbit and stability

• Beam envelope and projected emittance (slits)

• Invariant envelope matching (Ti:Sapph laser)

Main steps phase 2:

• Complete RF conditioning S-band structures

• Beam-based alignment of magnets and RF structures

• Commissioning BPMs: beam orbit and stability

• Commissioning of transverse deflecting structures:

bunch length and slice emittance

• Optics-based measurement of projected emittance

• Invariant envelope matching for minimum projected

emittance at 250 MeV

Main steps phase 3:

• RF conditioning X-band structures

• Beam-based alignment of X-band structure

• Bunch compressor studies: compression vs. S- and

X-band phase, deflection angle (R56)

• Micro-bunching and coherent optical transition radia-

tion

6.2 Operating regimes

In addition to the commissioning phases dictated by the

installation schedule, we define operating regimes in

terms of bunch charge and beam energy. On the one

hand this is required by radiation protection, such that

radiation measurements can be performed every time

a new operating regime is entered. On the other hand,

a finite set of operating regimes facilitates comparisons

between measurements as well as to simulation. The

charge regimes are defined according to Table 2 as 10 pC

(low charge), 100 pC (medium charge) and 200 pC (high

charge). For the energy, there are just two regimes, low

energy (gun only, up to 8 MeV), and high energy (full

linac, up to 280 MeV).

6.3 Software applications and data storage

Closely linkt to commissioning is the development of

software applications needed to perform the activities

listed above. Here the aim is to provide a set of tools

which will allow regular operators to execute the proce-

dures and measurements needed to operate the ma-

chine.

While simple manipulations of device readings are han-

dled within the low-level EPICS toolkit (see 7.1.2), more

advanced data treatment, in particular image processing,

is mainly done within the MATLAB computational frame-

work. A library for image processing has been developed

and is used by all applications based on screen images

(e.g., emittance measurements). Images and other

data related to specific measurements are stored in the

Nexus file format, a common standard in the X-ray, neu-

tron and muon community. Nexus itself is based on hdf5

(hierarchical data format 5).

Figure 37: Location of the temporary

shield inside the tunnel.

6 Commissioning and operation program


The control system plays a central role in all large fa-

cilities. It provides the connection between hardware

and the operators, it creates the environment that allows

physicists to do their measurements, and it archives

data to allow comparison with simulations retrospec-

tively. Guidelines for the design of the control system

are reliability, ease of use and expandability. Beyond the

pure functionality, the maintainability of the whole system

and the portability to new developments in computer

science are basic requirements.

The hardware and software platforms, boot and archive

services, configuration management and logistics devel-

oped for the other PSI accelerators shall be used as far

as possible. The control system for the 250 MeV Injector

will benefit from existing control systems equipment

standards and know-how of the other PSI accelerators

whenever possible. Many applications can be based on

the experience gained with SLS (Swiss Light Source) and

the LEG gun test facility. Developments needed for the

SwissFEL are to be accomplished in a manner that meets

the project schedule, budget, and performance needs,

with consideration for knowledge transfer – internally

and to the wider accelerator community.

7.1 Requirements

A wide range of user groups, like physicists and opera-

tors, will require information and support from the

control system. Therefore, on one hand, the control

system has to be designed to allow modularity and pos-

sibility of integration for new systems and programs to

ensure support of physicists. On the other hand, the

system should be as standardized and easy to use as

possible to support operators that have to care for all

PSI accelerators.

The 250 MeV Injector is foreseen to be used as the in-

jector in the final SwissFEL accelerator. That means that

the key concepts of the control system should take this

into account. But at the same time, the 250 MeV Injec-

tor provides the possibility to develop, test, and optimize

components for the use in the SwissFEL facility.

The control system shall be implemented in such a way

that it can be incrementally expanded and upgraded as

the project proceeds. The control system must be able

to be extended to new hardware and software technol-

ogy during the project phase and later.

7 Controls


The control system has to be laid out in a way that it will

scale up to the final scope of the facility without com-

promising performance. This includes the capability to

cope with a pulse repetition rate of 100 Hz, even if this

may not be needed for the 250 MeV Injector. Handling

issues that might be necessary only for the SwissFEL

like pulse rate limiting, restricted data rates (to operator

displays, for example), pulse identification and time

stamping and so on shall be considered in the design.

7.2 Control System Structure and Architecture

EPICS

The Control System will use the EPICS (Experimental

Physics and Industrial Control System) toolkit. EPICS

has been successfully applied at PSI in the SLS and in

several similar large projects around the world (for ex-

ample LCLS at SLAC). Due to its collaborative nature,

using EPICS enables us to take advantage of work done

at other laboratories. Using a standard software toolkit

will allow us to make best use of in-house know-how and

to consolidate technical support services.

Industry is supplying equipment with EPICS support and

there are companies that can be contracted to provide

EPICS support and applications when appropriate. This

is a considerable advantage over having to integrate

non-standard applications into the control system.

The EPICS toolkit includes:

• Network based client server model implementing Chan-

nel Access as network protocol

• Distributed process variable (PV) database that contain

a process image of the machine parameters

• EPICS extensions like Alarm Handler, Archiver and GUI

builder (medm) already known by operators

• Available Channel Access interfaces to common pro-

gramming languages (for example to C, C++, Java, Tcl/

Tk, Matlab and many more)

The general layout of an EPICS system is shown in Figure

38. The consoles in the control rooms run EPICS client

applications that connect over Ethernet network to the

distributed EPICS servers (called Input Output Controller)

which control the hardware. In addition central services

like the archiver are running on dedicated servers lo-

cated in a server room. In all of these computers, cus-

tomized programs requested by physicists can be imple-

mented and integrated into EPICS.

Software Maintenance and Distribution

One goal of the control system of the 250 MeV Injector

is to establish processes for easy distribution and main-

tenance of software for the SwissFEL accelerator. The

SwissFEL control system will be roughly five to ten times

bigger in terms of the number of controls components

needed. Therefore, automation should be introduced

where ever possible.

Figure 38: General EPICS system layout.

7 Controls


The general way of software distribution and installation

is already established at the SLS. Further development

might be needed to improve flexibility in order to integrate

new hard- and software.

The basic information needed for the automated software

installation is stored in relational databases. To make

best use of in-house knowledge the environment pro-

vided by the IT department, namely Oracle databases,

will be used.

Controls Hardware

For network components, file servers and PCs the 250

MeV Injector will use hardware based on PSI standards

defined by IT department.

As far as reasonable possible hardware used at other

PSI accelerators should be preferred to take advantage

of existing expertise and cheaper stock keeping. Such

hardware includes:

• VME crates and boards running the VxWorks Operating

System (from Wind River): PSI has adopted the VME64x

standard as its front-end hardware platform for the

accelerator facilities. It shall also be used for the 250

MeV Injector. A rich set of modules like digital and

analogue I/O as well as motor controller and encoder

cards is supported.

• PLCs type Siemens S7: For autonomous subsystems

and applications where high reliability is required but

the speed requirements are lower, PLCs will be used.

Typical applications are for example vacuum interlock,

component protection systems and so on. The applica-

tions will be not be developed by the controls section,

but have to be integrated into the control system for

visualization and data archiving.

• motor driver stages developed at PSI

• digital power supply controller developed at PSI

Special requirements may be solved by using embedded

controllers (for example FPGA-based or microcontrollers).

Examples of such systems may include geographically

isolated devices, systems that need the computing speed

of a dedicated CPU, or entirely new hardware components

which are not yet supported by any PSI standard hard-

ware. However, the embedded controllers shall be inte-

grated into the EPICS framework and the software

maintenance and distribution processes.

Network Layout

The network layout of the 250 MeV Injector is espe-

cially designed in view of the SwissFEL requirements.

There, the number of network ports needed is too high

to fit into a class C subnet, which is standard at PSI.

Therefore, the Injector will be used to develop, imple-

ment, and test the necessary procedures and programs

to connect different subnets along the accelerator.

The planed network layout is schematically shown in

Figure 39. Every circle in the sketch except "PSI Network"

represents a class C network with 254 addresses. The

connections of the different subnets of the EPICS net-

work are controlled by Channel Access Gateways to

minimize the network traffic. The control room is the

central network to minimize the number of network bor-

ders for measurements and programs running in the

control room.

Figure 39: Planned Network layout.

7.3 Special Subsystems

Timing System

The timing system provides synchronized event distribu-

tion on a 10 ns time scale. This allows measurements

and parameter settings to be triggered on defined time

distances from each other. For example the beam current

measurement can be triggered with a defined delay to

the electron bunch creation at the gun.

Furthermore, the timing system provides a bunch number

consisting of 52 bits. This allows for 252 bunch numbers

(4.5·1015 ), which will be unambiguous over a period of

time exceeding the lifetime of the SwissFEL at a repeti-

tion rate of 400 Hz.

7 Controls


The general functionality of the bunch numbering has

already been tested at the gun test facility. Further tests

will be done with the measurements planned in the injec-

tor to ensure easy usability of the system. The stable

and reproducible synchronization of the gun laser and

the RF phase will be an important point of investigation,

especially in view of stable user operation at the Swiss-

FEL.

The timing system is based on the SLS timing system

which will be adapted and extended to the requirements

of a linear accelerator with repetition rates up to 100

Hz. This might be needed only later in the SwissFEL ac-

celerator but should be tested as far as reasonably

possible in the 250 MeV Injector.

Interlock Systems

The interlock system could be divided into three main

parts which differ in their speed requirements, complex-

ity and protection goal:

1. PLC based Beam Loss Interlock System for radiation

protection

2. Beam Permit System for equipment protection, based

on the design of the Run Permit System used at the

HIPA accelerator at PSI. This system is not needed

for the 250 MeV Injector but is a test system for the

Swiss FEL

3. Beam Quality Verification System as part of the EPICS

environment, which will be designed and realized in

the course of first experiences derived from the 250

MeV Injector

Only the first two described systems are safety relevant.

They are independent of the EPICS environment in the

sense that they could work without EPICS. EPICS can be

used in these systems to visualize the situation and

archive the data.

In addition (and not described here) there is the vacuum

system with a dedicated interlock system that disables

the electron gun in case of bad vacuum, the personal

safety system (PSA) which is part of the radiation protec-

tion concept and local routines that supervise single

components like RF structures.

7 Controls


8.1 Introduction

For the SwissFEL injector test facility a new building [58]

in the PSI-West-area is to be built in order to test the

gun performance and beam line system, including a

beam dump for electron energies up to 0.3 GeV. This

document provides estimates of the radiation protection

relevant parameters, such as the lateral shielding thick-

ness and the beam dump dimensions and related shield-

ing thicknesses, used as guideline for the civil engineer

planning in 2009. The calculations are based on the

“SHIELD11 computer code” that will first be shortly in-

troduced.

8.2 The SHIELD11 computer code

The SHIELD11 code [59] has been developed at Stanford

Linear Accelerator Center (SLAC) to perform shielding

analyses in the vicinity of a high energy electron accel-

erator. It uses analytic expressions for the production

and attenuation of neutrons and photons from electron

beam loss at “thick targets” such as beam dumps or

collimators. The five major electron radiation loss com-

ponents considered are:

• Giant-Resonance-Neutrons (GRN): Those neutrons with

energy range between 0.1 MeV and 20 MeV are photo-

produced in the core of the shower.

• High-Energy-Neutrons (HEN): Neutrons with energy

above 100 MeV are initiated by high energy photons

in the hadronic cascade. The production angle is in

the forward direction.

• Mid-Energy-Neutrons (MID): Neutrons having energies

between GRN and HEN, being produced by quasi-

deuteron reactions.

• Direct Gamma (GAMD): Photons leaving the core with

energies between 0.1 MeV and 20 MeV.

• Indirect Gamma (GAMI): Photons and directly ionizing

particles created during slowing down of the HEN.

8.3 Radiation protection boundary condition

According to the conceptual design, the area outside

the accelerator bunker, though still inside the building,

will be accessible for persons not occupationally exposed

to ionizing radiation. In that sense the bunker shielding

has to be dimensioned to keep the ambient dose

equivalent rate below the limit of 0.125 μSv/h (the sum

of both, the neutron- and γ-dose rate).

8 Radiation Safety and Protection System


8.4 Shielding calculations

The operational conditions of the accelerator system to

be built are as follows. The final 300 MeV electron beam

will be generated by pulse operation with 10 Hz pulse

frequency and 40 psec pulse width with a peak current

of 5 A. The resulting charge per pulse is 200 pC and the

total beam power is 0.6 watt. The beam will finally be

dumped at the end of the beam line.

In Figure 40, the dose rate distribution for the five elec-

tron radiation loss contributions per beam loss of 1 watt,

at a distance of 1 m from the iron target (length 30.5 cm,

radius 5.1 cm) without any shielding are shown with

respect to the beam forward direction. The dose rate

maximum is located at a production angle ranging be-

tween 70 and 90 degrees. The major contribution found

in the direction of 90 degrees (this defines the thickness

of the lateral shielding) is mainly due to the direct pho-

tons from the electromagnetic shower with an energy

range <20 MeV. The second main contribution is due to

the Giant-Resonance neutrons, having energies between

0.1 MeV and 20 MeV.

Figure 40: Dose rate distribution for the five main electron loss

contributions, for a 1 watt beam-loss and a 0.3 MeV electron

beam hitting the iron target, as a function of the production angle

with respect to the beam forward direction, at distance of 1 m to

the target. Abbreviations used: NGES = total neutron does rate,

GAMGES = total γ-dose rate, TOTDOS = total dose rate

8.5 Lateral shielding

For the lateral shield calculation a shortest distance of

1.5 m between centre of beam and inner surface of the

shielding is given according to the ground plan drawing

of the injector building. The shielding wall is assumed

to consist of normal concrete (ρ = 2.35 g/cm3). Figure

42 shows the calculated shielding thickness distribution,

as a function of a beam loss range between 0.025 and

0.1 watt, for the three chosen electron beam energies

of 0.15 GeV (0.15 GeV is the threshold energy for the

HEN production in SHIELD11), 0.2 GeV and 0.3 GeV

respectively. The energy dependence of shielding thick-

ness is small and ranges from 5 cm to less than 10 cm

from the lowest to the highest beam losses considered.

An energy loss of about 5 % with respect to the total

beam power may be assumed [60] which is 0.03 Watt,

arising from some misalignment of the beam line. The

shielding thickness required in order to be below the

limits of the ambient dose rate, is between 114 cm and

118 cm, increasing with increasing beam loss. The mis-

steered beam may strike structures such as collimators

or iron yokes of the magnets, as well as some parts of

the beam vacuum tube itself. As already mentioned, the

SHIELD11 code is based on a “thick target” data ap-

proximation. In that sense, the thicknesses calculated

for the lateral shielding certainly will express losses in

collimators or other massive components. The inner

surface of a beam pipe (thickness 3 mm) may be hit by

an electron beam from grazing incidence (0.1 °C) to about

1 °C. The resulting effective thickness of the obstacle

varies between 1.7 m and 17 cm, corresponding to about

100 and 10 radiation lengths in iron respectively. That

is to say, an increasing deflection angle reduces target

thickness in terms of radiation length. The neutron yield

for iron for 1 GeV electrons increases by a factor of about

12 [61], when the target thickness varies from 2 up to

10 radiation lengths. It may be concluded that the lat-

eral shielding thicknesses shown are conservative,

considering interactions between electrons and the beam

tube.

Figure 41: Concrete shielding thickness as a function of beam

loss for several electron beam energies with the boundary condi-

tion of 0.125 μSv/h ambient dose equivalent rates outside the

shielding wall (area of public access).



Nevertheless, considering the standard concrete-shield-

ing-block thickness of 0.5 m and 1.0 m normally used

at PSI, the results indicate demand for a special solution.

8.6 Shielding of the silo roof

In a first attempt, the situation on the silo bunker may

be regarded as similar to the hall-floor. Allowing public

access during beam operation leads to the same shield-

ing thicknesses as for the lateral direction, presuming

a distance of 1.5 m between the beam line and the inner

surface of the concrete roof. If we consider the roof to

be a controlled zone of type 0 for example, an ambient

dose equivalent rate limit < 0 μSv/h would be tolerable

for unrestricted access of occupationally exposed per-

sons. In that case a roof thickness of 0.5 m would be

sufficient to hold the limit for 0.03 watt beam loss at

initial electron energy of 0.3 GeV. However the crane

driver should also be treated as a person occupation-

ally exposed to radiation (dosimeter wearing obligatory).

8.7 Shielding of the 0.3 GeV electron beam dump

The whole of the 0.6 watt 0.3 GeV electron beam will be

dumped at the end of the test-beam line. The shielding

calculations are based on a distance of 1.5 m between

the downstream end of beam dump and the inner surface

of the concrete wall and also laterally to the beam dump.

The limit for the ambient equivalent dose rate outside

the shielding is 0.125 μSv/h (area of public access).

The beam dump core is expected to consist of steel. It

can be shielded laterally and downstream with a layer

of lead. In Table 20 several possible beam dump con-

figurations are shown. Commonly the remaining γ-dose

rate is about 2 % compared to the neutron generated

one. A pure steel core target requires the longest dump

and the thickest shielding wall respectively. Additional

lead shielding in the forward direction allows for a reduc-

tion of the thickness of both, the steel core dump as

well as the concrete shielding wall. The activation po-

tential of the two metals is to be considered in the range

of disposal.

The thickness of the lateral concrete shielding for the

given beam dump lengths and geometries is affected

only by the additional lateral lead shielding. In Table 21

the results for three different lead thicknesses are

shown: 150 cm of concrete is adequate for a 15 cm

lateral lead cladding applied to the steel core beam

dump.



Length of beam dump iron (cm)

Addition lead downstream (cm)

Addition lead lateral (cm)

Thickness concrete wall downstream (cm)

γ-dose rate (μSv/h)

n-dose rate (μSv/h)

total-dose rate(μSv/h)

132 0 10 200 0.025 0.09 0.11

50 20 10 200 0.02 0.075 0.095

132 36 10 150 0.017 0.08 0.097

50 44 10 150 0.016 0.074 0.09

Addition lead lateral [cm]

Resulting thick-ness of lateral concrete wall [cm]

γ-dose rate Thickness concrete wall downstream (cm)

γ-dose rate (μSv/h)

n-dose rate (μSv/h)

total-dose rate(μSv/h)

10 153 0.013 0.081 0.094 0.09 0.11

12 150 0.013 0.081 0.094 0.075 0.095

15 145 0.012 0.083 0.095 0.08 0.097

Table 20: Beam dump shielding in beam forward direction at production angle 0 degree.

Table 21: Lateral beam dump shielding at production angle 90 degree.



[1] Editor R. Ganter, SwissFEL CDR, to be published.

[2] C. Limborg et al., “RF design of the LCLS gun”,LCLS

TN-05-03.

[3] T. Schietinger et al., measuring and modeling at the

PSI-XFEL 500-kV Low Emittance Electron Source,

Proceedings of the LINAC 2008, Vancoover, Canada

(2008).

[4] Y. Kim et al., Beam Diagnostic and RF Systems

Requirements for the SwisFEL Facility, Proceedings

DIPAC09.

[5] B.E. Carlsten, Nucl. Instrum. Methods Phys. Res., Sect.

A285, 313–319 (1989).

[6] C. Wang et al, Criteria for emittance compensation in

high-brightness photoinjectors, PHYSICAL REVIEW

SPECIAL TOPICS – ACCELERATORS AND BEAMS 10,

104201 (2007).

[7] Handbook of Accelerator Physics and Engineering, Sec.

2.4.4, p. 104, Eds. A. W. Chao, M. Tigner, World

Schientific Publishing Co. Pte. Ltd., Singapore, ISBN

9810235005 / 91802388584 (2002).

[8] Recent Advances and Novel Ideas for High Brightness

Electron Beam Production Based on Photo-Injectors

M. Ferrario, M. Boscolo, V. Fusco, C. Vaccarezza,

C. Ronsivalle, J.B. Rosenzweig, L. Serafini, LFN internal

report SPARC-BD-03/003/LNF–03/06 (P), INFN,

Frascati, Italy (2003).

[9] Envelope analysis of intense relativistic quasilaminar

beams in r f photoinjectors: A theory of emittance

compensation. L. Serafini, J.B. Rosenzweig, Phys. Rev.

E., 55 (6), p. 7565 (1997).

[10] M. Altarelli et al, European Free-Electron Laser Technical

Design Report (page 98).

[11] SLAC–R–593, Linac Coherent Light Source, Conceptual

design report (paragraph 6.1).

[12] Y. Ding et al, SLAC-PUB-13525, January 2009.

[13] R. Ganter et al, Commissioning of a Diode/RF Photogun

Combination, Proc. FEL 2009, Liverpool, UK.

[14] R. Bossard et all, CLIC note 297 and PS/RF note 95-25

(1995).

[15] R. Bossart, M. Dehler, DESIGN OF A RF GUN FOR HEAVY

BEAM LOADING, EPAC 96 proceedings.

[16] L. Stingelin, FEL-SU84-001-Power_test-CTF-Gun,

SwissFEL internal report.

[17] R. Losito et al, “The PHIN Photoinjector for the CTF3

drive beam”, Proceeding EPAC 06.

[18] M. Dehler et al., Phys. Rev.ST accelerators and beams

12, 2009.

[19] D. Alesini et al., RF deflector design and measurements

for the longitudinal and transverse phase space

characterizationat SPARC, Nucl. Instrum. Meth. A 568,

488-502, 2006.

[20] G. D’Auria, private communications.

[21] L. Doolittle et al., Digital Low-Level RF Control Using

Non-IQ Sampling, Proceedings LINAC06, Knoxville,

Tennessee USA.

[22] Y. Otake. Construction of XFEL/Spring8, Proceedings

of FEL08, Gyeongju, Korea.

[23] S. Hunziker et al., Toward an Ultra-Stable Reference

Distribution for the Nex PSI 250 MeV Injector,

proceeding DIPAC09, Basel Switzerland.

9 References

9 References


[24] A. Winter, private communication, 2007.

[25] E. Vogel, W. Koprek, P. Pucyk, in FPGA based RF field

control at the photo cathode RF gun of the DESY

vacuum ultraviolet free electron laser, Proceedings of

the EPAC Conference, Edinburgh, Scotland, 2006.

[26] R. Akre, D. Dowell, P. Emma, J. Frisch, B. Hong,

K. Kotturi, P. Krejcik, J. Wu, in LCLS LLRF Upgrades to

the SLAC LINAC, Proceedings of the PAC Conference,

Albuquerque, New Mexico, USA, 2007.

[27] C. Rivetta, R. Akre, K. Kotturi, in Operational Experience

with LCLS RF systems, LCLS DOE Review, July 11,

2007.

[28] V. Schlott et al., “Performance of the Digital BPM

System for the Swiss Light Source”, Proc. DIPAC’01,

Grenoble, France, May 2001.

[29] M. Böge et al., “User Operation and Upgrades of the

Fast Orbit Feedback at the SLS”, Proc. PAC’05,

Knoxville, USA, May 2005.

[30] M. Dehler et al., “State of the SLS Multi Bunch

Feedbacks“, Proc. APAC’07, Indore, India, Feb. 2007.

[31] A. Andersson et al., “Recent Results from the Electron

Beam Profile Monitor at the Swiss Light Source”, Proc.

DIPAC’07, Venice, Italy, May 2007.

[32] D. Sütterlin, “Single Shot Electron Bunch Length

Measurements with a Spatial Electro-Optical Auto-

Correlation Interferometer using Coherent Transition

Radiation at the 100 MeV SLS Pre-Injector LINAC”,

PHD-thesis, Diss ETH Nr. 16668, ETH-Zürich 2006.

[33] V. Schlott et al., “THz Diagnostic for the Femto Bunch

Slicing Project at the Swiss Light Source”, Proc.

EPAC’06, Edinburgh, UK, June 2006.

[34] R. Ischebeck et al., ”Screen Monitor Design for the

SwissFEL”, Proc. DIPAC 2009, Basel, Switzerland, May

2009.

[35] A. Citterio et al., “Design of a Resonant Stripline Beam

Position Pickup for the 250 MeV PSI XFEL Test Injector”,

Proc. DIPAC 2009, Basel, Switzerland, May 2009.

[36] B. Keil et al., ” Application of a 5 GSPS Analogue Ring

Sampling Chip for Low-cost Single-shot BPM Systems”,

Proc. EPAC 2008, Genoa, June 2008.

[37] See: http://www.bergoz.com/products/In-Flange.CT/

In-Flange.CT.html

[38] B. Steffen et al., “A Compact Single Shot Electro-Optical

Bunch Length Monitor for the SwissFEL”, DIPAC 2009,

Basel, Switzerland, May 2009.

[39] F. Müller et al., “Ytterbium Fiber Laser for Electro-Optical

Pulse Length Measurements at the SwissFEL”, DIPAC

2009, Basel, Switzerland, May 2009.

[40] F. Löhl et al., ”A Sub-50fs Bunch Arrival Time Monitor

System for FLASH”, Proc. DIPAC’07, Venice, Italy, May

2007.

[41] http://www.iapla.unibe.ch/

[42] B.J. Claessens et al., Phys. Rev. Letters 95, 164801

(2005).

[43] www.fastlite.com

[44] B.W. Van Wonterghem et al, Proceedings of the

Conference on Laser Coherence Control, SPIE 1870

(1993), p 64.

[45] www.quantel.fr

[46] C. Hauri et all, Intrinsic Emittance Reduction of an

Electron Beam from Metal Photocathodes, Phys. Rev.

Lett. 104, 234802 (2010).

[47] L. Canova et al, “Ultrashor tpulse generation with the

Mazzler active spectral broadening and the XPW pulse

shortening technique, Conference on Lasers an Electro-

Optics, OSA Technical Digest 2008, paper JTuA39.

[48] LCLS Conceptual design report, ssrl.slac.stanford.

edu/lcls/cdr.

[49] P. Wiegand, "SLS SR girder, vibration and modal

analysis tests", SLS-TME-TA-2000-0153.

[50] Engineering Study Report on the Magnetic Elements

and Girders for Swiss Light Source, BudkerInstitute,

Novosibirsk, October 1997.

[51] Y. Kim et al., Low Thermal Emittance Measurements

at the SwissFEL Low Emittance Gun test Facility,

Proceeding of FEL 08, Gyeongju, Korea.

[52] D.H. Dowell et al, Quantum efficiency and thermal

emittance in metal photocathodes, Phys. Rev. Special

Topics, Accelerator and beams, 12, 074201, 2009.

[53] Y. Kim et al., Start to end simulations of the PSI 250

MeV injector test facility, proceedings of EPAC08,

Genoa, Italy.

[54] L. Serafini, J.B. Rosenzweig, Phys. Rev. E 55.

[55] M. Ferrario et al., SLAC-PUB-8400 (1997) 7565.

[56] J. Welch et al, LCLS compressor dipole field quality

and beam measurements, Proceedings of FEL08,

Gyeongju, Korea, 2008.

[57] www.vectorfields.com

[58] Memorandum “LEG Project New Building Shielding”

Ch. Gough 29.05.06, PSI.

[59] The SHIELD11 Computer Code W.R. Nelson,

T.M. Jenkins, SLAC-R-737, UC-414.

[60] Private communication Ch. Gough, PSI.

[61] “Giant dipole resonance neutron yields produced by

electron as a function of target material and thickness”,

X. Mao et al., SLAC-PUB-6628, 1996.

9 References


Published by

Paul Scherrer Institut

Coordination

Marco Pedrozzi

English Proofreading

Trevor Dury

Design and Layout

Irma Herzog

Photographs

© Paul Scherrer Institut

Printing


Available from


Communications Services

5232 Villigen PSI, Switzerland

Phone +41 (0)56 310 21 11

www.psi.ch

PSI public relations

[email protected]

Copying is welcomed, provided

the source is acknowledged

and an archive copy sent to PSI.


July 2010

Paul Scherrer Institut, 5232 Villigen PSI, Switzerland

Tel. +41 (0)56 310 21 11, Fax +41 (0)56 310 21 99www.psi.ch

Documents

SwissFEL Injector Conceptual Design Report · in July 2010 during the assembly phase. PSI – SwissFEL Injector Conceptual Design Report 3 ntroduction1 I 4 2 Mission of the test facility